Model-Informed Approaches and Innovative Clinical Trial Design for Adeno-Associated Viral Vector-Based Gene Therapy Product Development: A White Paper.

The promise of viral vector-based gene therapy (GT) as a transformative paradigm for treating severely debilitating and life-threatening diseases is slowly coming to fruition with the recent approval of several drug products. However, they have a unique mechanism of action often necessitating a tortuous clinical development plan. Expertise in such complex therapeutic modality is still fairly limited in this emerging class of adeno-associated virus (AAV) vector-based gene therapies. Because of the irreversible mode of action and incomplete understanding of genotype-phenotype relationship and disease progression in rare diseases careful considerations should be given to GT product's benefit-risk profile. In particular, special attention needs to be paid to safe dose selection, reliable dose exposure response (including clinically relevant endpoints), or creative approaches in study design targeting small patient populations during clinical development. We believe that quantitative tools encompassed within model-informed drug development (MIDD) framework fits quite well in the development of such novel therapies, as they enable us to benefit from the totality of data approach in order to support dose selection as well as optimize clinical trial designs, end point selection, and patient enrichment. In this thought leadership paper, we provide our collective experiences, identify challenges, and suggest areas of improvement in applications of modeling and innovative trial design in development of AAV-based GT products and reflect on the challenges and opportunities for incorporating MIDD tools and more in rational development of these products.

Current practices for QSP model assessment: an IQ consortium survey.

Quantitative Systems Pharmacology (QSP) modeling is increasingly applied in the pharmaceutical industry to influence decision making across a wide range of stages from early discovery to clinical development to post-marketing activities. Development of standards for how these models are constructed, assessed, and communicated is of active interest to the modeling community and regulators but is complicated by the wide variability in the structures and intended uses of the underlying models and the diverse expertise of QSP modelers. With this in mind, the IQ Consortium conducted a survey across the pharmaceutical/biotech industry to understand current practices for QSP modeling. This article presents the survey results and provides insights into current practices and methods used by QSP practitioners based on model type and the intended use at various stages of drug development. The survey also highlights key areas for future development including better integration with statistical methods, standardization of approaches towards virtual populations, and increased use of QSP models for late-stage clinical development and regulatory submissions.

A microfluidic system that replicates pharmacokinetic (PK) profiles in vitro improves prediction of in vivo efficacy in preclinical models.

Test compounds used on in vitro model systems are conventionally delivered to cell culture wells as fixed concentration bolus doses; however, this poorly replicates the pharmacokinetic (PK) concentration changes seen in vivo and reduces the predictive value of the data. Herein, proof-of-concept experiments were performed using a novel microfluidic device, the Microformulator, which allows in vivo like PK profiles to be applied to cells cultured in microtiter plates and facilitates the investigation of the impact of PK on biological responses. We demonstrate the utility of the device in its ability to reproduce in vivo PK profiles of different oncology compounds over multiweek experiments, both as monotherapy and drug combinations, comparing the effects on tumour cell efficacy in vitro with efficacy seen in in vivo xenograft models. In the first example, an ERK1/2 inhibitor was tested using fixed bolus dosing and Microformulator-replicated PK profiles, in 2 cell lines with different in vivo sensitivities. The Microformulator-replicated PK profiles were able to discriminate between cell line sensitivities, unlike the conventional fixed bolus dosing. In a second study, murine in vivo PK profiles of multiple Poly(ADP-Ribose) Polymerase 1/2 (PARP) and DNA-dependent protein kinase (DNA-PK) inhibitor combinations were replicated in a FaDu cell line resulting in a reduction in cell growth in vitro with similar rank ordering to the in vivo xenograft model. Additional PK/efficacy insight into theoretical changes to drug exposure profiles was gained by using the Microformulator to expose FaDu cells to the DNA-PK inhibitor for different target coverage levels and periods of time. We demonstrate that the Microformulator enables incorporating PK exposures into cellular assays to improve in vitro-in vivo translation understanding for early therapeutic insight.

Quantitative Evaluation of Dendritic Nanoparticles in Mice: Biodistribution Dynamics and Downstream Tumor Efficacy Outcomes.

A physiologically based pharmacokinetic model was developed to describe the tissue distribution kinetics of a dendritic nanoparticle and its conjugated active pharmaceutical ingredient (API) in plasma, liver, spleen, and tumors. Tumor growth data from MV-4-11 tumor-bearing mice were incorporated to investigate the exposure/efficacy relationship. The nanoparticle demonstrated improved antitumor activity compared to the conventional API formulation, owing to the extended released API concentrations at the site of action. Model simulations further enabled the identification of critical parameters that influence API exposure in tumors and downstream efficacy outcomes upon nanoparticle administration. The model was utilized to explore a range of dosing schedules and their effect on tumor growth kinetics, demonstrating the improved antitumor activity of nanoparticles with less frequent dosing compared to the same dose of naked APIs in conventional formulations.

Design and optimisation of dendrimer-conjugated Bcl-2/xL inhibitor, AZD0466, with improved therapeutic index for cancer therapy.

Dual Bcl-2/Bcl-xL inhibitors are expected to deliver therapeutic benefit in many haematological and solid malignancies, however, their use is limited by tolerability issues. AZD4320, a potent dual Bcl-2/Bcl-xL inhibitor, has shown good efficacy however had dose limiting cardiovascular toxicity in preclinical species, coupled with challenging physicochemical properties, which prevented its clinical development. Here, we describe the design and development of AZD0466, a drug-dendrimer conjugate, where AZD4320 is chemically conjugated to a PEGylated poly-lysine dendrimer. Mathematical modelling was employed to determine the optimal release rate of the drug from the dendrimer for maximal therapeutic index in terms of preclinical anti-tumour efficacy and cardiovascular tolerability. The optimised candidate is shown to be efficacious and better tolerated in preclinical models compared with AZD4320 alone. The AZD4320-dendrimer conjugate (AZD0466) identified, through mathematical modelling, has resulted in an improved therapeutic index and thus enabled progression of this promising dual Bcl-2/Bcl-xL inhibitor into clinical development.

AZD0364 Is a Potent and Selective ERK1/2 Inhibitor That Enhances Antitumor Activity in KRAS-Mutant Tumor Models when Combined with the MEK Inhibitor, Selumetinib.

The RAS-regulated RAF-MEK1/2-ERK1/2 (RAS/MAPK) signaling pathway is a major driver in oncogenesis and is frequently dysregulated in human cancers, primarily by mutations in BRAF or RAS genes. The clinical benefit of inhibitors of this pathway as single agents has only been realized in BRAF-mutant melanoma, with limited effect of single-agent pathway inhibitors in KRAS-mutant tumors. Combined inhibition of multiple nodes within this pathway, such as MEK1/2 and ERK1/2, may be necessary to effectively suppress pathway signaling in KRAS-mutant tumors and achieve meaningful clinical benefit. Here, we report the discovery and characterization of AZD0364, a novel, reversible, ATP-competitive ERK1/2 inhibitor with high potency and kinase selectivity. In vitro, AZD0364 treatment resulted in inhibition of proximal and distal biomarkers and reduced proliferation in sensitive BRAF-mutant and KRAS-mutant cell lines. In multiple in vivo xenograft models, AZD0364 showed dose- and time-dependent modulation of ERK1/2-dependent signaling biomarkers resulting in tumor regression in sensitive BRAF- and KRAS-mutant xenografts. We demonstrate that AZD0364 in combination with the MEK1/2 inhibitor, selumetinib (AZD6244 and ARRY142886), enhances efficacy in KRAS-mutant preclinical models that are moderately sensitive or resistant to MEK1/2 inhibition. This combination results in deeper and more durable suppression of the RAS/MAPK signaling pathway that is not achievable with single-agent treatment. The AZD0364 and selumetinib combination also results in significant tumor regressions in multiple KRAS-mutant xenograft models. The combination of ERK1/2 and MEK1/2 inhibition thereby represents a viable clinical approach to target KRAS-mutant tumors.

AZD4320, A Dual Inhibitor of Bcl-2 and Bcl-xL, Induces Tumor Regression in Hematologic Cancer Models without Dose-limiting Thrombocytopenia.

PURPOSE: Targeting Bcl-2 family members upregulated in multiple cancers has emerged as an important area of cancer therapeutics. While venetoclax, a Bcl-2-selective inhibitor, has had success in the clinic, another family member, Bcl-xL, has also emerged as an important target and as a mechanism of resistance. Therefore, we developed a dual Bcl-2/Bcl-xL inhibitor that broadens the therapeutic activity while minimizing Bcl-xL-mediated thrombocytopenia. EXPERIMENTAL DESIGN: We used structure-based chemistry to design a small-molecule inhibitor of Bcl-2 and Bcl-xL and assessed the activity against in vitro cell lines, patient samples, and in vivo models. We applied pharmacokinetic/pharmacodynamic (PK/PD) modeling to integrate our understanding of on-target activity of the dual inhibitor in tumors and platelets across dose levels and over time. RESULTS: We discovered AZD4320, which has nanomolar affinity for Bcl-2 and Bcl-xL, and mechanistically drives cell death through the mitochondrial apoptotic pathway. AZD4320 demonstrates activity in both Bcl-2- and Bcl-xL-dependent hematologic cancer cell lines and enhanced activity in acute myeloid leukemia (AML) patient samples compared with the Bcl-2-selective agent venetoclax. A single intravenous bolus dose of AZD4320 induces tumor regression with transient thrombocytopenia, which recovers in less than a week, suggesting a clinical weekly schedule would enable targeting of Bcl-2/Bcl-xL-dependent tumors without incurring dose-limiting thrombocytopenia. AZD4320 demonstrates monotherapy activity in patient-derived AML and venetoclax-resistant xenograft models. CONCLUSIONS: AZD4320 is a potent molecule with manageable thrombocytopenia risk to explore the utility of a dual Bcl-2/Bcl-xL inhibitor across a broad range of tumor types with dysregulation of Bcl-2 prosurvival proteins.

Multiscale Model Identifies Improved Schedule for Treatment of Acute Myeloid Leukemia In Vitro With the Mcl-1 Inhibitor AZD5991.

Anticancer efficacy is driven not only by dose but also by frequency and duration of treatment. We describe a multiscale model combining cell cycle, cellular heterogeneity of B-cell lymphoma 2 family proteins, and pharmacology of AZD5991, a potent small-molecule inhibitor of myeloid cell leukemia 1 (Mcl-1). The model was calibrated using in vitro viability data for the MV-4-11 acute myeloid leukemia cell line under continuous incubation for 72 hours at concentrations of 0.03-30 µM. Using a virtual screen, we identified two schedules as having significantly different predicted efficacy and showed experimentally that a "short" schedule (treating cells for 6 of 24 hours) is significantly better able to maintain the rate of cell kill during treatment than a "long" schedule (18 of 24 hours). This work suggests that resistance can be driven by heterogeneity in protein expression of Mcl-1 alone without requiring mutation or resistant subclones and demonstrates the utility of mathematical models in efficiently identifying regimens for experimental exploration.

Discovery of a Potent and Selective Oral Inhibitor of ERK1/2 (AZD0364) That Is Efficacious in Both Monotherapy and Combination Therapy in Models of Nonsmall Cell Lung Cancer (NSCLC).

The RAS/MAPK pathway is a major driver of oncogenesis and is dysregulated in approximately 30% of human cancers, primarily by mutations in the BRAF or RAS genes. The extracellular-signal-regulated kinases (ERK1 and ERK2) serve as central nodes within this pathway. The feasibility of targeting the RAS/MAPK pathway has been demonstrated by the clinical responses observed through the use of BRAF and MEK inhibitors in BRAF V600E/K metastatic melanoma; however, resistance frequently develops. Importantly, ERK1/2 inhibition may have clinical utility in overcoming acquired resistance to RAF and MEK inhibitors, where RAS/MAPK pathway reactivation has occurred, such as relapsed BRAF V600E/K melanoma. We describe our structure-based design approach leading to the discovery of AZD0364, a potent and selective inhibitor of ERK1 and ERK2. AZD0364 exhibits high cellular potency (IC50 = 6 nM) as well as excellent physicochemical and absorption, distribution, metabolism, and excretion (ADME) properties and has demonstrated encouraging antitumor activity in preclinical models.

Discovery of Mcl-1-specific inhibitor AZD5991 and preclinical activity in multiple myeloma and acute myeloid leukemia.

Mcl-1 is a member of the Bcl-2 family of proteins that promotes cell survival by preventing induction of apoptosis in many cancers. High expression of Mcl-1 causes tumorigenesis and resistance to anticancer therapies highlighting the potential of Mcl-1 inhibitors as anticancer drugs. Here, we describe AZD5991, a rationally designed macrocyclic molecule with high selectivity and affinity for Mcl-1 currently in clinical development. Our studies demonstrate that AZD5991 binds directly to Mcl-1 and induces rapid apoptosis in cancer cells, most notably myeloma and acute myeloid leukemia, by activating the Bak-dependent mitochondrial apoptotic pathway. AZD5991 shows potent antitumor activity in vivo with complete tumor regression in several models of multiple myeloma and acute myeloid leukemia after a single tolerated dose as monotherapy or in combination with bortezomib or venetoclax. Based on these promising data, a Phase I clinical trial has been launched for evaluation of AZD5991 in patients with hematological malignancies (NCT03218683).

Mixture dynamics: Combination therapy in oncology.

In recent years combination therapies have become increasingly popular in most therapeutic areas. We present a qualitative and quantitative approach and elucidate some of the challenges and solutions to a more optimal therapy. For tumor growth this involves the study of semi-mechanistic cell-growth/kill models with multiple sites of action. We introduce such models and analyze their dynamic properties using simulations and mathematical analysis. This is done for two specific case studies, one involving a single compound and one a combination of two compounds. We generalize the notion of Tumor Static Concentration to cases when two compounds are involved and develop a graphical method for determining the optimal combination of the two compounds, using ideas akin to those used in studies employing isobolograms. In studying the dynamics of the second case study we focus, not only on the different concentrations, but also on the different dosing regimens and pharmacokinetics of the two compounds.

AZD1208, a potent and selective pan-Pim kinase inhibitor, demonstrates efficacy in preclinical models of acute myeloid leukemia.

Upregulation of Pim kinases is observed in several types of leukemias and lymphomas. Pim-1, -2, and -3 promote cell proliferation and survival downstream of cytokine and growth factor signaling pathways. AZD1208 is a potent, highly selective, and orally available Pim kinase inhibitor that effectively inhibits all three isoforms at <5 nM or <150 nM in enzyme and cell assays, respectively. AZD1208 inhibited the growth of 5 of 14 acute myeloid leukemia (AML) cell lines tested, and sensitivity correlates with Pim-1 expression and STAT5 activation. AZD1208 causes cell cycle arrest and apoptosis in MOLM-16 cells, accompanied by a dose-dependent reduction in phosphorylation of Bcl-2 antagonist of cell death, 4EBP1, p70S6K, and S6, as well as increases in cleaved caspase 3 and p27. Inhibition of p4EBP1 and p-p70S6K and suppression of translation are the most representative effects of Pim inhibition in sensitive AML cell lines. AZD1208 inhibits the growth of MOLM-16 and KG-1a xenograft tumors in vivo with a clear pharmacodynamic-pharmacokinetic relationship. AZD1208 also potently inhibits colony growth and Pim signaling substrates in primary AML cells from bone marrow that are Flt3 wild-type or Flt3 internal tandem duplication mutant. These results underscore the therapeutic potential of Pim kinase inhibition for the treatment of AML.

FuncBase: a resource for quantitative gene function annotation.

SUMMARY: Computational gene function prediction can serve to focus experimental resources on high-priority experimental tasks. FuncBase is a web resource for viewing quantitative machine learning-based gene function annotations. Quantitative annotations of genes, including fungal and mammalian genes, with Gene Ontology terms are accompanied by a community feedback system. Evidence underlying function annotations is shown. For example, a custom Cytoscape viewer shows functional linkage graphs relevant to the gene or function of interest. FuncBase provides links to external resources, and may be accessed directly or via links from species-specific databases. AVAILABILITY: FuncBase as well as all underlying data and annotations are freely available via http://func.med.harvard.edu/

A critical assessment of Mus musculus gene function prediction using integrated genomic evidence.

BACKGROUND: Several years after sequencing the human genome and the mouse genome, much remains to be discovered about the functions of most human and mouse genes. Computational prediction of gene function promises to help focus limited experimental resources on the most likely hypotheses. Several algorithms using diverse genomic data have been applied to this task in model organisms; however, the performance of such approaches in mammals has not yet been evaluated. RESULTS: In this study, a standardized collection of mouse functional genomic data was assembled; nine bioinformatics teams used this data set to independently train classifiers and generate predictions of function, as defined by Gene Ontology (GO) terms, for 21,603 mouse genes; and the best performing submissions were combined in a single set of predictions. We identified strengths and weaknesses of current functional genomic data sets and compared the performance of function prediction algorithms. This analysis inferred functions for 76% of mouse genes, including 5,000 currently uncharacterized genes. At a recall rate of 20%, a unified set of predictions averaged 41% precision, with 26% of GO terms achieving a precision better than 90%. CONCLUSION: We performed a systematic evaluation of diverse, independently developed computational approaches for predicting gene function from heterogeneous data sources in mammals. The results show that currently available data for mammals allows predictions with both breadth and accuracy. Importantly, many highly novel predictions emerge for the 38% of mouse genes that remain uncharacterized.

Combining guilt-by-association and guilt-by-profiling to predict Saccharomyces cerevisiae gene function.

BACKGROUND: Learning the function of genes is a major goal of computational genomics. Methods for inferring gene function have typically fallen into two categories: 'guilt-by-profiling', which exploits correlation between function and other gene characteristics; and 'guilt-by-association', which transfers function from one gene to another via biological relationships. RESULTS: We have developed a strategy ('Funckenstein') that performs guilt-by-profiling and guilt-by-association and combines the results. Using a benchmark set of functional categories and input data for protein-coding genes in Saccharomyces cerevisiae, Funckenstein was compared with a previous combined strategy. Subsequently, we applied Funckenstein to 2,455 Gene Ontology terms. In the process, we developed 2,455 guilt-by-profiling classifiers based on 8,848 gene characteristics and 12 functional linkage graphs based on 23 biological relationships. CONCLUSION: Funckenstein outperforms a previous combined strategy using a common benchmark dataset. The combination of 'guilt-by-profiling' and 'guilt-by-association' gave significant improvement over the component classifiers, showing the greatest synergy for the most specific functions. Performance was evaluated by cross-validation and by literature examination of the top-scoring novel predictions. These quantitative predictions should help prioritize experimental study of yeast gene functions.

An en masse phenotype and function prediction system for Mus musculus.

BACKGROUND: Individual researchers are struggling to keep up with the accelerating emergence of high-throughput biological data, and to extract information that relates to their specific questions. Integration of accumulated evidence should permit researchers to form fewer - and more accurate - hypotheses for further study through experimentation. RESULTS: Here a method previously used to predict Gene Ontology (GO) terms for Saccharomyces cerevisiae (Tian et al.: Combining guilt-by-association and guilt-by-profiling to predict Saccharomyces cerevisiae gene function. Genome Biol 2008, 9(Suppl 1):S7) is applied to predict GO terms and phenotypes for 21,603 Mus musculus genes, using a diverse collection of integrated data sources (including expression, interaction, and sequence-based data). This combined 'guilt-by-profiling' and 'guilt-by-association' approach optimizes the combination of two inference methodologies. Predictions at all levels of confidence are evaluated by examining genes not used in training, and top predictions are examined manually using available literature and knowledge base resources. CONCLUSION: We assigned a confidence score to each gene/term combination. The results provided high prediction performance, with nearly every GO term achieving greater than 40% precision at 1% recall. Among the 36 novel predictions for GO terms and 40 for phenotypes that were studied manually, >80% and >40%, respectively, were identified as accurate. We also illustrate that a combination of 'guilt-by-profiling' and 'guilt-by-association' outperforms either approach alone in their application to M. musculus.

Chipper: discovering transcription-factor targets from chromatin immunoprecipitation microarrays using variance stabilization.

Chromatin immunoprecipitation combined with microarray technology (Chip2) allows genome-wide determination of protein-DNA binding sites. The current standard method for analyzing Chip2 data requires additional control experiments that are subject to systematic error. We developed methods to assess significance using variance stabilization, learning error-model parameters without external control experiments. The method was validated experimentally, shows greater sensitivity than the current standard method, and incorporates false-discovery rate analysis. The corresponding software ('Chipper') is freely available. The method described here should help reveal an organism's transcription-regulatory 'wiring diagram'.

Towards a proteome-scale map of the human protein-protein interaction network.

Systematic mapping of protein-protein interactions, or 'interactome' mapping, was initiated in model organisms, starting with defined biological processes and then expanding to the scale of the proteome. Although far from complete, such maps have revealed global topological and dynamic features of interactome networks that relate to known biological properties, suggesting that a human interactome map will provide insight into development and disease mechanisms at a systems level. Here we describe an initial version of a proteome-scale map of human binary protein-protein interactions. Using a stringent, high-throughput yeast two-hybrid system, we tested pairwise interactions among the products of approximately 8,100 currently available Gateway-cloned open reading frames and detected approximately 2,800 interactions. This data set, called CCSB-HI1, has a verification rate of approximately 78% as revealed by an independent co-affinity purification assay, and correlates significantly with other biological attributes. The CCSB-HI1 data set increases by approximately 70% the set of available binary interactions within the tested space and reveals more than 300 new connections to over 100 disease-associated proteins. This work represents an important step towards a systematic and comprehensive human interactome project.

Genomewide identification of Sko1 target promoters reveals a regulatory network that operates in response to osmotic stress in Saccharomyces cerevisiae.

In Saccharomyces cerevisiae, the ATF/CREB transcription factor Sko1 (Acr1) regulates the expression of genes induced by osmotic stress under the control of the high osmolarity glycerol (HOG) mitogen-activated protein kinase pathway. By combining chromatin immunoprecipitation and microarrays containing essentially all intergenic regions, we estimate that yeast cells contain approximately 40 Sko1 target promoters in vivo; 20 Sko1 target promoters were validated by direct analysis of individual loci. The ATF/CREB consensus sequence is not statistically overrepresented in confirmed Sko1 target promoters, although some sites are evolutionarily conserved among related yeast species, suggesting that they are functionally important in vivo. These observations suggest that Sko1 association in vivo is affected by factors beyond the protein-DNA interaction defined in vitro. Sko1 binds a number of promoters for genes directly involved in defense functions that relieve osmotic stress. In addition, Sko1 binds to the promoters of genes encoding transcription factors, including Msn2, Mot3, Rox1, Mga1, and Gat2. Stress-induced expression of MSN2, MOT3, and MGA1 is diminished in sko1 mutant cells, while transcriptional regulation of ROX1 seems to be unaffected. Lastly, Sko1 targets PTP3, which encodes a phosphatase that negatively regulates Hog1 kinase activity, and Sko1 is required for osmotic induction of PTP3 expression. Taken together our results suggest that Sko1 operates a transcriptional network upon osmotic stress, which involves other specific transcription factors and a phosphatase that regulates the key component of the signal transduction pathway.

Predicting protein complex membership using probabilistic network reliability.

Evidence for specific protein-protein interactions is increasingly available from both small- and large-scale studies, and can be viewed as a network. It has previously been noted that errors are frequent among large-scale studies, and that error frequency depends on the large-scale method used. Despite knowledge of the error-prone nature of interaction evidence, edges (connections) in this network are typically viewed as either present or absent. However, use of a probabilistic network that considers quantity and quality of supporting evidence should improve inference derived from protein networks. Here we demonstrate inference of membership in a partially known protein complex by using a probabilistic network model and an algorithm previously used to evaluate reliability in communication networks.