Advanced drug delivery reviews最新文献

{"title":"Overcoming barriers and shaping the future: Challenges and innovations in nucleic acid therapies for Glioblastoma","authors":"Alaa Zam ,&nbsp;Nadia Rouatbi ,&nbsp;Adam A. Walters ,&nbsp;Khuloud T. Al-Jamal","doi":"10.1016/j.addr.2025.115759","DOIUrl":"10.1016/j.addr.2025.115759","url":null,"abstract":"<div><div>Glioblastoma (GBM) is the most aggressive and treatment-resistant primary brain tumor in adults. Conventional therapies offer limited benefit due to the tumor's heterogeneity, invasive nature, and the presence of the blood–brain barrier, which restricts therapeutic access. Nucleic acid (NA)-based therapies, including small interfering RNA, microRNA, antisense oligonucleotides, splice-switching oligonucleotides, and CRISPR-based systems, have emerged as promising tools to modulate oncogenic pathways and overcome resistance mechanisms at the genetic level. However, effective delivery remains the primary challenge in translating these therapies into clinical success. This review examines the current landscape of NA-based strategies for GBM, with a focus on innovative delivery systems designed to navigate biological barriers and enhance therapeutic precision. We highlight clinical progress made with nanocarrier platforms such as liposomes, lipid nanoparticles, and exosome-based systems, and evaluate their safety, specificity, and delivery efficiency. Additionally, we discuss the most promising preclinical advances, including multifunctional, targeted, and stimuli-responsive carriers, that demonstrate strong potential for clinical translation. Our analysis underscores that the therapeutic efficacy of NA approaches in GBM is inseparable from the sophistication of their delivery platforms. Moving forward, the integration of rationally designed carriers with gene-targeted payloads holds the key to unlocking the full potential of precision medicine in GBM.</div></div>","PeriodicalId":7254,"journal":{"name":"Advanced drug delivery reviews","volume":"229 ","pages":"Article 115759"},"PeriodicalIF":17.6,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145717923","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Small data, big challenges: Machine- and deep-learning strategies for data-limited drug discovery","authors":"Nazreen Pallikkavaliyaveetil ,&nbsp;Sriram Chandrasekaran","doi":"10.1016/j.addr.2025.115762","DOIUrl":"10.1016/j.addr.2025.115762","url":null,"abstract":"<div><div>A critical bottleneck limiting the potential of Machine Learning (ML) and Deep Learning (DL) models within the drug discovery and development (DDD) pipeline is the scarcity of high-quality experimental data. Limited data is not an anomaly but an inherent characteristic of the DDD process. Significant financial costs, time, and confidentiality concerns limit the scale of available datasets. Applying standard ML and DL algorithms directly to these small datasets presents substantial challenges. Traditional ML models remain constrained by their dependence on handcrafted features and limited ability to capture complex biological relationships. In contrast, DL algorithms that assume data abundance are prone to overfitting and poor generalization when trained on small datasets. The small data problem thus represents a fundamental constraint that shapes the practical utility and trustworthiness of AI applications in DDD. While prior reviews have surveyed the broad landscape of AI and ML in drug discovery, a significant gap exists concerning the small data challenge across the DDD pipeline. Addressing this challenge requires adapting DL methods that typically assume data abundance, while also extending traditional ML approaches that, although well-suited to small data, remain limited in their representational capacity. This review addresses this gap by surveying key drug discovery tasks, highlighting the prevalence of limited data, and synthesizing both traditional ML methods and advanced DL strategies tailored to these contexts. By integrating methodological advances with task-specific applications, the review outlines current approaches and identifies opportunities for advancing robust, interpretable, and generalizable AI in drug discovery.</div></div>","PeriodicalId":7254,"journal":{"name":"Advanced drug delivery reviews","volume":"229 ","pages":"Article 115762"},"PeriodicalIF":17.6,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145771264","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"GMP-compliant manufacturing of iPSC-derived therapeutic cell products: Technologies, applications, risks and limitations","authors":"Alexandra Haase,&nbsp;Arjang Ruhparwar,&nbsp;Ulrich Martin","doi":"10.1016/j.addr.2025.115744","DOIUrl":"10.1016/j.addr.2025.115744","url":null,"abstract":"<div><div>The development of induced pluripotent stem cells (iPSCs) has transformed the field of regenerative medicine. However, to use iPSCs for therapeutic applications, iPSC-based products must be produced under Good Manufacturing Practice (GMP) conditions. This process involves reprogramming somatic cells, characterizing and banking iPSC lines, introducing therapeutic transgenes if necessary, and scaling up cell expansion and differentiation for clinical use. This review provides an overview of the relevant regulatory authorities and relevant regulations in the US, Europe, and Japan. It also discusses the current challenges and opportunities in producing GMP-compliant iPSCs. These challenges include the need for defined culture media, as well as developing all the required GMP-compliant processes, such as reprogramming, establishing iPSC clones, and manufacturing processes that lead to the final advanced therapy medicinal product (ATMP). For autologous products in particular, this can be complicated by cell line-specific variation of proliferation velocity and differentiation biases. The review also discusses attempts to develop automated closed systems. It emphasizes the importance of ensuring the sterility, identity, (<em>epi</em>)genetic integrity, and functionality of the final cell products to guarantee the safety and the efficacy of iPSC-based therapies. However, the need for reproducibility, rigorous quality control and safety requirements has resulted in high regulatory hurdles and extremely high costs, which often prevent the initiation of clinical trials. Overcoming these challenges will enable iPSCs to play an integral role in future medicine and offer new treatment options for various diseases.</div></div>","PeriodicalId":7254,"journal":{"name":"Advanced drug delivery reviews","volume":"229 ","pages":"Article 115744"},"PeriodicalIF":17.6,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145657300","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Exploiting colloidal drug aggregation for drug delivery: From promise to prediction using computational tools","authors":"Kai V. Slaughter ,&nbsp;Xiang Olivia Li ,&nbsp;Molly S. Shoichet","doi":"10.1016/j.addr.2025.115758","DOIUrl":"10.1016/j.addr.2025.115758","url":null,"abstract":"<div><div>Colloidal drug aggregates are amorphous nanoparticles formed by the self-assembly of hydrophobic small molecule drugs. They can be leveraged as drug-rich nanoparticle formulations for drug delivery. However, it is difficult to predict which drugs will form colloidal aggregates, which stabilizers will be effective, and what the <em>in vivo</em> fate of the nanoparticles will be. These challenges can be addressed, in part, with computational tools including artificial intelligence such as machine learning. Molecular dynamics simulations have been used to improve our understanding of the intermolecular forces that govern the assembly of colloidal drug aggregates. Several predictive tools exist to identify aggregators, but these are typically used to eliminate aggregators from screening libraries rather than design drug delivery formulations. Colloidal drug aggregates require stabilizers to prevent particle growth and precipitation. Computational analyses have been used to predict which colloidal drug aggregators can be stabilized by a particular small molecule excipient and to identify drug-stabilizer pairs. Successful stabilization has enabled colloidal drug aggregate evaluation for applications such as nanomedicine and sustained release. Additionally, certain colloid-forming drugs can be useful for co-delivery of nucleic acids. In future studies, computational tools can be developed to predict the biological activity of colloidal drug aggregates, building upon other approaches currently used for lipid nanoparticles and other modalities. Ultimately, leveraging computational strategies to improve the design of colloidal drug aggregates can help realize the potential of this high drug-loading delivery platform.</div></div>","PeriodicalId":7254,"journal":{"name":"Advanced drug delivery reviews","volume":"229 ","pages":"Article 115758"},"PeriodicalIF":17.6,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145717893","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The growing impact of machine learning on drug formulation science","authors":"Zeqing Bao ,&nbsp;Steven Huynh ,&nbsp;Frantz Le Devedec ,&nbsp;Jo Nguyen ,&nbsp;Christine Allen","doi":"10.1016/j.addr.2025.115761","DOIUrl":"10.1016/j.addr.2025.115761","url":null,"abstract":"<div><div>Machine learning (ML) has increasingly been adopted in drug formulation science to support more efficient, data-driven drug development strategies. A growing number of studies have demonstrated the promise of ML tools across various aspects of drug formulation science, including both preformulation studies and formulation optimization. Building on these foundational efforts, more advanced data collection and ML techniques are now being integrated, driving innovation and expanding the scope of ML applications in the field. To better understand the trend of breakthroughs in this area, this review examines relevant works published in the past decade, identifying key trends, core applications, and emerging techniques in ML-driven drug delivery. Representative studies are highlighted as examples to illustrate the evolving landscape and practical implementations of these technologies. Furthermore, this review explores forward-looking perspectives, highlighting the convergence of ML with the increasing openness of regulatory bodies, the integration of organoid models, and the advancement of experimental automation.</div></div>","PeriodicalId":7254,"journal":{"name":"Advanced drug delivery reviews","volume":"229 ","pages":"Article 115761"},"PeriodicalIF":17.6,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145731924","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Quality by Computational Design: Harnessing AI to Advance Rational Drug Development","authors":"Nannan Wang ,&nbsp;Hao Zhong ,&nbsp;Ping Xiong ,&nbsp;Jinying Zhu ,&nbsp;Defang Ouyang","doi":"10.1016/j.addr.2025.115764","DOIUrl":"10.1016/j.addr.2025.115764","url":null,"abstract":"<div><div>The pharmaceutical Quality by Design (QbD) principle aims to reduce risk and improve efficiency across drug development lifecycle. However, QbD was originally established in an era preceding the widespread adoption of artificial intelligence (AI) and did not fully capture the potential of computational pharmaceutics. This gap is particularly pronounced in complex drug product development, where conventional QbD relies on empirical knowledge and labor-intensive experimentation. As a result, it struggles to accommodate multi-modal and multi-scale variables, and lacks sufficient flexibility, dynamic optimization capabilities, and the ability to perform clinically oriented inverse design. In recent years, advances in computational pharmaceutics have provided a new methodological foundation for drug development. In this context, we propose a novel paradigm, termed Quality by Computational Design (QbCD), which integrates computational pharmaceutics within the QbD framework to achieve mechanism-based and clinically guided formulation design. We first define the concept of QbCD, outline its essential components, implementation steps, and methodological strengths, and discuss relevant regulatory considerations. Building on this, we propose a practical QbCD implementation guideline to strengthen model credibility and ensure regulatory compliance. Subsequently, to establish the methodological foundation and demonstrate practical feasibility, we present the core techniques of QbCD, including AI, physical modeling, and <em>in vivo</em> modeling, and examine their applications across various stages of drug development. To further illustrate the practicality of QbCD, two representative cases are presented: a QbCD-enabled virtual development workflow for amorphous solid dispersions and a real-world implementation of QbCD in designing long-acting <em>in situ</em> gel injectables. Finally, we discuss future perspectives for QbCD, focusing on bridging the data gap, advancing methodological innovations, enhancing model credibility and regulatory compliance, and fostering a supportive scientific culture and ecosystem in computational pharmaceutics. These efforts aim to promote a more intelligent, efficient, and clinically aligned paradigm for rational drug development.</div></div>","PeriodicalId":7254,"journal":{"name":"Advanced drug delivery reviews","volume":"229 ","pages":"Article 115764"},"PeriodicalIF":17.6,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145796068","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Microbial heterogeneity-mediated treatment evasion and the potential for engineered live biotherapeutic products","authors":"Eli G. Cytrynbaum ,&nbsp;Megan N. McClean","doi":"10.1016/j.addr.2025.115740","DOIUrl":"10.1016/j.addr.2025.115740","url":null,"abstract":"<div><div>Microbial populations exhibit both genetic and non-genetic heterogeneity. In the clinical context, this heterogeneity is of concern as it provides subsets of microbial populations with enhanced immune evasion abilities and antimicrobial resistance. Fungal pathogens are of special concern as fungal diseases and antifungal resistance are increasing worldwide and similarities between eukaryotic cells make it challenging to identify targets that are toxic to fungi without also harming the human host. Engineered live biotherapeutic products (eLBPs) could provide an alternative and complementary approach to manipulating and treating heterogeneous fungal populations due to their potential to provide localized delivery to the affected site, continuous long-term treatment, environmental sensing, and delivery of therapeutics specific to virulent or drug-resistant organisms. However, the development of eLBPs targeting fungi remains limited.</div><div>This review assesses our current understanding of genetic and non-genetic microbial heterogeneity and how this impacts treatment strategies particularly for pathogenic fungi. We focus on <em>Candida</em> yeasts, specifically <em>Candida albicans</em>, as <em>Candida</em> species are the most common opportunistic fungal pathogens. We review the current scope and potential of eLBPs to address heterogeneous and rising fungal infections.</div></div>","PeriodicalId":7254,"journal":{"name":"Advanced drug delivery reviews","volume":"229 ","pages":"Article 115740"},"PeriodicalIF":17.6,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145611319","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Guiding design and performance of nonviral nucleic acid delivery vehicles via machine learning","authors":"John A. Hutchinson ,&nbsp;Sidharth Panda ,&nbsp;Plinio D. Rosales ,&nbsp;Janey P. Sowada ,&nbsp;Miles S. Willis ,&nbsp;Michael C. Leyden ,&nbsp;Prodromos Daoutidis ,&nbsp;Theresa M. Reineke","doi":"10.1016/j.addr.2025.115739","DOIUrl":"10.1016/j.addr.2025.115739","url":null,"abstract":"<div><div>Machine Learning (ML) techniques have enabled the advancement of many technologies throughout the pharmaceutical industry, especially for drug discovery. One of the most rapidly growing technologies within the pharmaceutical space is gene therapy, with twenty six FDA-approved genetic medicines and over three thousand treatments currently undergoing clinical trials. A key challenge in the successful employment of gene therapy is effective nucleic acid delivery, and nonviral delivery vectors provide a cost-effective and highly customizable solution to this challenge. However, the vast design space also poses a large challenge for traditional development, which relies heavily on iterative trial-and-error and costly <em>in vivo</em> and <em>in vitro</em> experiments. This review identifies key ML techniques and discusses how these approaches have been utilized to improve the design of nonviral nucleic acid delivery vehicles. ML has the potential to radically transform the design space for nucleic acid therapies, like it has already done in drug discovery and drug formulations. This potential is being realized in research and has already led to the advent of several commercial enterprises seeking to build full end-to-end platforms for rapidly decreasing development time for new gene therapies.</div></div>","PeriodicalId":7254,"journal":{"name":"Advanced drug delivery reviews","volume":"229 ","pages":"Article 115739"},"PeriodicalIF":17.6,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145583533","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Engineering tumor spatial heterogeneity in vitro","authors":"Changchong Chen ,&nbsp;Zixuan Zhao ,&nbsp;Dong Hua Seah ,&nbsp;Kenny Zhuoran Wu ,&nbsp;Senthilkumar Mohanaselvi ,&nbsp;Eliza Li Shan Fong","doi":"10.1016/j.addr.2025.115757","DOIUrl":"10.1016/j.addr.2025.115757","url":null,"abstract":"<div><div>Spatial heterogeneity is a fundamental feature of the tumor microenvironment, characterized by structured variations in cellular composition, phenotypic states, extracellular matrix (ECM) organization, and biochemical and biophysical gradients. These spatial patterns shape tumor evolution, modulate immune infiltration, and underlie resistance to therapy. Advances in spatial transcriptomics and multiplex imaging have revealed dynamic and region-specific niches, such as hypoxic cores, immune-excluded zones, and fibroblast-dense invasive fronts, that correlate with clinical outcomes. However, most in vitro models fail to capture this architectural complexity. Recent engineering technologies, including 3D bioprinting, organoid assembloids, organ-on-a-chip systems, and ECM-mimetic scaffolds, now enable controlled reconstruction of tumor spatial organization and microregional heterogeneity. These technologies allow integration of patient-derived cells, tunable matrix environments, and spatially defined signaling to mimic in vivo pathophysiology. When integrated with spatial transcriptomics and proteomics, these models enable mechanistic exploration of microregional tumor biology, evaluation of therapeutic responses, and investigation of immunotherapy resistance. This review integrates our current understanding of spatial heterogeneity in cancer with enabling engineering strategies to guide future developments in tumor biology and therapeutic innovation.</div></div>","PeriodicalId":7254,"journal":{"name":"Advanced drug delivery reviews","volume":"229 ","pages":"Article 115757"},"PeriodicalIF":17.6,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145731927","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Emerging nanoparticle-based therapies for pancreatic cancer: Current clinical landscape","authors":"Ainara Salgado-Pascual ,&nbsp;Sara Zalba ,&nbsp;Juan José Lasarte ,&nbsp;Maria J. Garrido","doi":"10.1016/j.addr.2025.115760","DOIUrl":"10.1016/j.addr.2025.115760","url":null,"abstract":"<div><div>Pancreatic cancer, particularly ductal adenocarcinoma (PDAC) is one of the most aggressive and lethal subtypes due to late diagnosis, the absence of early biomarkers, and the presence of a complex tumor microenvironment (TME). This TME is characterized by a dense desmoplastic stroma, hypovascularization, immunosuppression, and an acidic extracellular pH. All of these factors hinder the delivery and efficacy of conventional therapies, especially in advanced stages.</div><div>Nanoparticles (NPs), including liposomes, polymeric micelles, albumin-bound particles and lipid nanoparticles, have emerged as a promising tool for overcoming TME barriers, and enhance drug delivery in tumor while minimizing systemic toxicity. NPs can exploit mechanisms such as the Enhanced Permeability and Retention (EPR) effect and active targeting. Clinically approved NPs such as Nab-Paclitaxel and liposomal Irinotecan have demonstrated improved pharmacokinetics and therapeutic benefits in PDAC. Furthermore, ongoing clinical trials are exploring novel NP-based strategies such as gene delivery, radiosensitization, immunomodulation and ferroptosis induction. Despite these promising advances, significant challenges remain, including poor tumor penetration, heterogeneity in EPR, and immune recognition of NPs leading to their clearance from bloodstream before reaching the tumor. Innovative solutions such as biomimetic coatings, stimuli-responsive systems and personalized formulations, are being evaluated to enhance NP performance. Standardization of NP characterization and data reporting are essential to accelerating clinical translation. The integration of artificial intelligence and machine learning could further optimize NP design and patient stratification. Overall, nanotechnology represents a crucial frontier of research for developing more effective and personalized pancreatic cancer treatments.</div></div>","PeriodicalId":7254,"journal":{"name":"Advanced drug delivery reviews","volume":"229 ","pages":"Article 115760"},"PeriodicalIF":17.6,"publicationDate":"2026-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145731034","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}