Progress in RNA-based treatments: recent discoveries, medical use, and future outlook

Messenger RNA (mRNA) vaccines represent a transformative approach in vaccinology, exemplified by their rapid development and deployment during the COVID-19 pandemic. Early research on mRNA vaccines dates back to the 1990s, but progress was initially hindered by instability issues, immunogenicity, and inefficient delivery systems (Wolff et al., 1990; Hoerr et al., 2000) Figure 1. However, breakthroughs in nucleotide modifications, specifically the incorporation of modified nucleosides such as pseudouridine, significantly improved the stability and reduced the immune activation of mRNA, facilitating effective clinical application (Karikó et al., 2005a; Karikó and Weissman, 2007). The mRNA vaccines developed by Pfizer-BioNTech and Moderna demonstrated unprecedented efficacy and rapid production timelines, marking a new era in vaccine technology (Polack et al., 2020; Baden et al., 2021). These vaccines function by delivering synthetically produced mRNA encoding the SARS-CoV-2 spike protein, inducing host cells to transiently produce viral antigens and stimulate robust immune responses without introducing an infectious agent (Jackson et al., 2020). Beyond infectious diseases, mRNA vaccines have also shown promising potential in oncology as personalized cancer vaccines, wherein patient-specific tumor antigens encoded by mRNA elicit tailored immune responses against cancer cells (Sahin and Türeci, 2018).

Small interfering RNAs (siRNAs) exploit the endogenous RNA interference (RNAi) pathway discovered initially in plants and later elucidated in animals, highlighting a conserved mechanism across species for gene silencing (Fire et al., 1998; Elbashir et al., 2001). Clinical validation of siRNA therapeutics was significantly advanced by the approval of Patisiran, which targets transthyretin mRNA to treat hereditary transthyretin-mediated amyloidosis (Adams et al., 2018). Similarly, Givosiran, an approved siRNA therapy for acute hepatic porphyria, underscores the versatility and therapeutic potential of siRNAs in treating metabolic disorders by precisely silencing pathogenic genes (Balwani et al., 2020). Despite their clinical successes, siRNA therapeutics continue to face challenges such as delivery efficiency, immune activation, and specificity, which ongoing research aims to overcome through advanced chemical modifications and novel delivery platforms such as lipid nanoparticles (LNPs) (Setten et al., 2019; Kulkarni et al., 2018).

Antisense oligonucleotides (ASOs) are short synthetic RNA or DNA molecules designed to hybridize specifically to complementary mRNA sequences, modulating their processing or translation. Initial discoveries in antisense technologies occurred in the late 1970s and early 1980s, paving the way for modern therapeutics (Zamecnik and Stephenson, 1978). Spinraza, an ASO approved for spinal muscular atrophy, alters the splicing of survival motor neuron 2 (SMN2) pre-mRNA to increase the production of functional SMN protein, significantly improving patient outcomes (Mercuri et al., 2018). Another no example, Eteplirsen, approved for Duchenne muscular dystrophy, employs exon-skipping technology to restore the dystrophin reading frame and produce functional dystrophin protein (Mendell et al., 2016). Continuous improvements in ASO design, stability, and delivery are critical to further expand their therapeutic applications across diverse genetic disorders (Crooke, 2017).

Emerging RNA therapeutic modalities, notably CRISPR-based RNA targeting, have opened novel avenues for precision medicine. The discovery of the CRISPR-Cas system as a bacterial adaptive immune mechanism has revolutionized genome editing technologies (Jinek et al., 2012). The CRISPR-Cas13 system, in particular, allows specific RNA targeting and cleavage without affecting DNA, offering transient and reversible gene expression modulation (Abudayyeh et al., 2017). This RNA-editing technology has demonstrated potential in targeting viral RNAs, correcting disease-associated transcripts, and modulating gene expression with high specificity and minimal off-target effects (Gootenberg et al., 2017). Ongoing research aims to optimize CRISPR-Cas13 efficacy, specificity, and delivery methods, further expanding its potential for diverse therapeutic applications in viral infections, genetic disorders, and beyond (Pickar-Oliver and Gersbach, 2019).

Lipid nanoparticles have become the most widely utilized platform for RNA delivery, particularly highlighted by their success in COVID-19 mRNA vaccines (Hou et al., 2021). The concept of lipid-based delivery dates back to studies in the 1980s exploring liposome-mediated gene transfer (Felgner et al., 1987). Early RNA delivery attempts suffered from rapid degradation by extracellular RNases and poor cellular uptake (Whitehead et al., 2009). The breakthrough came with the design of ionizable lipids that are neutral at physiological pH but become positively charged in acidic environments such as endosomes, facilitating endosomal escape and cytoplasmic RNA release (Cullis and Hope, 2017). Ionizable lipid formulations reduce toxicity and immunogenicity compared to permanently cationic lipids (Heyes et al., 2005). In addition to ionizable lipids, LNPs contain helper lipids, cholesterol, and polyethylene glycol (PEG)-lipids to optimize stability and circulation time (Sahay et al., 2020).

Clinically, LNPs enable high RNA encapsulation efficiency and favorable pharmacokinetics, leading to robust protein expression in target tissues (Kulkarni et al., 2018). The Pfizer-BioNTech and Moderna vaccines used LNPs with optimized ionizable lipids to safely deliver SARS-CoV-2 spike protein-encoding mRNA, illustrating the real-world success of this technology (Polack et al., 2020; Baden et al., 2021). Beyond vaccines, LNPs are employed for siRNA drugs such as Onpattro (Patisiran) for hereditary transthyretin amyloidosis, further validating their versatility (Adams et al., 2018).

However, challenges remain, including potential PEG immunogenicity leading to accelerated blood clearance upon repeated dosing and limitations in tissue targeting beyond the liver (Irvine et al., 2020). Recent research focuses on novel lipid chemistries, biodegradable ionizable lipids, and incorporation of targeting ligands to improve delivery specificity and reduce adverse effects (Lonez et al., 2014; Cheng et al., 2020).

Viral vectors, including adenoviruses, adeno-associated viruses (AAV), and lentiviruses, offer efficient gene delivery by exploiting natural viral infection mechanisms (Naldini, 2015). While predominantly used for DNA delivery, these vectors have also been adapted for RNA delivery in some contexts. Their advantages include high transduction efficiency and potential for long-term expression, crucial for certain therapeutic indications (Mingozzi and High, 2013). However, viral vectors face significant limitations: immunogenicity triggering neutralizing antibodies and inflammation, limited packaging capacity (particularly for AAVs), complex manufacturing, and safety concerns related to insertional mutagenesis (Thomas et al., 2003; Ginn et al., 2018). The immune response against viral vectors often restricts repeat dosing, posing challenges for chronic therapies (Shirley et al., 2020). Engineering less immunogenic viral capsids, use of serotype switching, and transient immunosuppression strategies are being explored to mitigate these challenges (Kotterman and Schaffer, 2014).

Polymeric nanoparticles represent one of the most versatile classes of RNA delivery vehicles due to their tunable chemistry, biodegradability, and potential for controlled release. Commonly studied polymers include poly (lactic-co-glycolic acid) (PLGA), polyethylenimine (PEI), poly (β-amino esters) (PBAEs), and dendrimers, each offering distinct physicochemical properties that can be tailored for nucleic acid encapsulation (Danhier et al., 2012; Verbaan et al., 2005). PLGA nanoparticles, already FDA-approved for other drug formulations, provide favorable safety profiles and sustained release kinetics, but often require surface modifications (e.g., PEGylation or ligand conjugation) to enhance cellular uptake and protect RNA cargo. In contrast, cationic polymers like PEI demonstrate high transfection efficiency through strong electrostatic interactions with RNA, yet are limited by dose-dependent cytotoxicity and variable in vivo performance (Mintzer and Simanek, 2009). To overcome these limitations, biodegradable or low-molecular-weight PEI derivatives and PBAEs have been engineered to balance transfection efficiency with reduced toxicity (Sunshine et al., 2012).

Emerging approaches also leverage stimuli-responsive polymers, which release RNA cargo in response to pH, redox state, or enzymatic activity within target tissues, improving precision and minimizing systemic side effects (Rosenblum et al., 2020). Hybrid systems that integrate polymeric and lipid components are gaining attention for combining the stability and functionalization capacity of polymers with the high encapsulation efficiency of lipids (Wang et al., 2018).

Beyond synthetic polymers, biological carriers such as exosomes and extracellular vesicles (EVs) offer natural RNA transport mechanisms, with intrinsic tissue tropism and low immunogenicity (El Andaloussi et al., 2013). Preclinical studies demonstrate successful delivery of mRNA and siRNA using engineered exosomes; however, reproducible large-scale manufacturing and standardized loading methods remain major hurdles (Lener et al., 2015; Armstrong and Stevens, 2018). Although still in early translational stages compared to lipid nanoparticles, polymeric and exosome-based delivery systems hold unique promise for extrahepatic targeting, repeat dosing, and crossing biological barriers such as the blood–brain barrier. Continued progress will likely depend on integrating polymer chemistry, bioengineering, and scalable manufacturing to achieve clinically viable platforms.

Chemical modification of RNA molecules has been pivotal in improving stability and reducing innate immune activation (Karikó et al., 2005b). Incorporation of modified nucleotides such as pseudouridine and 5-methylcytidine reduces recognition by Toll-like receptors and RIG-I, enhancing translational efficiency and reducing inflammation (Anderson et al., 2010; Svitkin et al., 2017). Concurrently, conjugation of RNA molecules or their delivery vehicles with targeting ligands has facilitated cell- and tissue-specific delivery, thereby increasing therapeutic indices and reducing systemic toxicity. Ligands such as N-acetylgalactosamine (GalNAc), aptamers, and antibodies enable selective binding to receptors on target cells, promoting efficient uptake primarily through receptor-mediated endocytosis (Nair et al., 2014; Whitehead et al., 2014). Notably, GalNAc conjugation has been extensively utilized for hepatocyte targeting via the asialoglycoprotein receptor (ASGPR), a strategy that underpins several clinically approved siRNA therapeutics for liver diseases.

The concept of RNA interference (RNAi) was first elucidated in Caenorhabditis elegans by Fire et al., in 1998, demonstrating potent, sequence-specific gene silencing via double-stranded RNA (Fire et al., 1998). This discovery catalyzed the exploration of synthetic small interfering RNAs (siRNAs) as therapeutics. Early delivery hurdles were addressed by chemical stabilization strategies (e.g., phosphorothioate backbones) and nanoparticle encapsulation to protect siRNAs from nuclease degradation (Whitehead et al., 2009). The first clinical trial of an siRNA drug, ALN-VSP, targeted liver cancer using LNP delivery, laying groundwork for RNAi therapeutics (Zimmermann et al., 2006). However, systemic administration raised challenges related to off-target effects and immune stimulation (Judge et al., 2006). The subsequent FDA approval of Patisiran in 2018 validated these early efforts by demonstrating safe and effective delivery of siRNA via LNPs to hepatocytes, with meaningful clinical improvement in hereditary transthyretin amyloidosis (Adams et al., 2018). Antisense oligonucleotides (ASOs), first described in the late 1970s by Zamecnik and Stephenson, also faced initial challenges in stability and delivery (Zamecnik and Stephenson, 1978). Through extensive chemical modification (2′-O-methyl, phosphorothioate) and optimization, ASOs like Nusinersen have become standard of care for spinal muscular atrophy, highlighting their clinical viability (Mercuri et al., 2018; Crooke, 2017).

The advent of mRNA vaccines represents a landmark in RNA therapeutic development, accelerated dramatically by the urgency of the COVID-19 pandemic. While mRNA technology had been under investigation for decades, its clinical translation was limited by RNA instability and immunogenicity (Wolff et al., 1990; Karikó et al., 2005b). Key innovations—such as nucleoside modifications that diminish innate immune recognition and lipid nanoparticle (LNP) delivery systems that enhance cellular uptake and protect mRNA—catalyzed breakthroughs (Cullis and Hope, 2017; Anderson et al., 2010). The Pfizer-BioNTech (BNT162b2) and Moderna (mRNA-1273) vaccines exemplify these advancements. Both vaccines utilize nucleoside-modified mRNA encoding the SARS-CoV-2 spike protein, delivered via optimized LNPs. Clinical trials demonstrated approximately 95% efficacy in preventing symptomatic COVID-19 with acceptable safety profiles (Polack et al., 2020; Baden et al., 2021). These trials were completed within months, a pace previously unparalleled in vaccine development.

Regulatory agencies leveraged Emergency Use Authorizations (EUAs) to expedite access, balancing rapid review with rigorous safety assessments (U.S. Food and Drug Administration (FDA), 2020). Post-marketing surveillance has provided real-world validation of vaccine safety and effectiveness, despite rare adverse events such as myocarditis, primarily in young males (Mevorach et al., 2021). This success story validates mRNA platforms for rapid response to emergent pathogens and establishes a blueprint for future vaccine development, including for cancer and other infectious diseases (Sahin and Türeci, 2018; Pardi et al., 2018).

RNA therapeutics challenge traditional regulatory paradigms, requiring adaptation to their unique manufacturing and safety profiles. Regulatory guidance documents, such as those from the FDA and EMA, emphasize stringent characterization of RNA integrity, purity, and the physicochemical properties of delivery systems, particularly lipid nanoparticles (FDA, 2018; European Medicines Agency (EMA), 2025).

Quality control for RNA therapeutics includes assessment of RNA sequence fidelity, capped structures, and chemical modifications, which impact stability and immunogenicity (Kauffman et al., 2016). Analytical methods such as next-generation sequencing (NGS), capillary electrophoresis, and mass spectrometry are routinely employed for lot release and stability studies. Nonclinical safety assessments focus on immunogenicity, off-target effects, and biodistribution, leveraging in vitro and in vivo models to predict human responses (Setten et al., 2019). Regulatory agencies increasingly encourage incorporation of pharmacogenomics and biomarker studies in clinical trials to identify responders and monitor adverse events.The rapid clinical development of COVID-19 mRNA vaccines underscored the utility of flexible regulatory pathways, including adaptive trial designs, rolling reviews, and conditional approvals (U.S. Food and Drug Administration (FDA), 2020). However, these accelerated frameworks also highlight the need for robust pharmacovigilance systems to monitor long-term safety and efficacy post-approval.

The inherent immunostimulatory properties of RNA pose significant safety considerations. Unmodified RNA activates innate immune receptors including Toll-like receptors (TLR3, TLR7, TLR8) and RIG-I-like receptors, potentially inducing inflammatory cytokines and systemic reactions (Hornung et al., 2006; Karikó et al., 2005a). Chemical modifications such as pseudouridine and 5-methylcytidine reduce these responses by altering recognition patterns (Anderson et al., 2010; Svitkin et al., 2017). Delivery systems can also provoke immune responses. PEGylated lipids, used to stabilize LNPs, have been associated with rare hypersensitivity reactions and accelerated blood clearance upon repeat dosing, necessitating ongoing optimization (Irvine et al., 2020).

Off-target gene silencing, a concern primarily for siRNA and ASO therapies, can result in unintended downregulation of genes leading to toxicity. Advanced bioinformatics, transcriptome-wide binding studies, and chemical modifications reduce this risk (Jackson et al., 2016; Setten et al., 2019). Additionally, rigorous clinical monitoring for organ toxicities, particularly hepatic and renal, is standard practice. Emerging gene-editing RNA therapies, such as CRISPR-Cas9 systems, require detailed off-target analysis to preempt genome instability or mutagenesis (Gillmore et al., 2021). Safety monitoring protocols incorporate next-generation sequencing to detect unintended edits, combined with long-term follow-up for genotoxicity.

A major policy gap exists in the regulation of n-of-1 or ultra-personalized RNA therapeutics. The rapid development of Milasen, an antisense oligonucleotide created for a single patient with Batten disease (Kim et al., 2019), underscores both the feasibility and the regulatory vacuum in such scenarios. Current FDA and EMA frameworks do not yet provide clear, standardized pathways for compassionate use or accelerated approval of bespoke RNA therapies, raising ethical questions around equity, cost, and long-term monitoring. Proposed frameworks include adaptive trial designs, patient registries for real-world data capture, and policy harmonization across jurisdictions to ensure that individualized therapies can be developed safely while remaining accessible.

The landscape of RNA therapeutics continues to expand with innovations such as self-amplifying RNA (saRNA), circular RNA (circRNA), and novel gene-editing approaches. saRNA leverages replicon technology to amplify intracellular RNA, allowing lower doses and prolonged protein expression, currently in early clinical development (Alameh et al., 2021). circRNAs offer enhanced stability due to their covalently closed loop structures, representing promising candidates for sustained protein production (Wesselhoeft et al., 2018).

CRISPR-based therapies are progressing rapidly, with in vivo genome editing trials for genetic disorders such as transthyretin amyloidosis demonstrating safety and efficacy (Gillmore et al., 2021). Delivery challenges remain significant for gene editors, driving research into improved lipid and viral vector platforms. Artificial intelligence and machine learning are increasingly integrated into RNA drug discovery and safety profiling, accelerating candidate optimization and enabling personalized therapeutics. Regulatory frameworks are anticipated to evolve in tandem with these advances, ensuring patient safety while promoting innovation.

Immunogenicity is among the foremost challenges confronting RNA therapeutics. Exogenous RNA molecules are recognized by innate immune receptors, including endosomal Toll-like receptors (TLR3, TLR7, TLR8) and cytosolic sensors such as RIG-I and MDA5, which detect pathogen-associated molecular patterns and trigger pro-inflammatory cascades (Hornung et al., 2006; Kawai and Akira, 2008). The resulting activation of type I interferon pathways can lead to systemic inflammation, cytokine release syndrome, and attenuation of therapeutic efficacy due to rapid RNA clearance (Karikó et al., 2005b; Robbins et al., 2020).

Chemical modification of RNA nucleosides, notably the incorporation of pseudouridine, 5-methylcytidine, and N1-methyl-pseudouridine, has proven effective in mitigating immunogenicity by reducing recognition by these pattern recognition receptors, thus enhancing translational capacity and reducing inflammatory side effects (Karikó et al., 2005a; Anderson et al., 2010; Svitkin et al., 2017). However, even chemically modified RNAs are not entirely inert, especially with repeated administrations, where immune memory can exacerbate responses (Kaczmarek et al., 2017). Additionally, delivery vehicles contribute to immunogenicity. Polyethylene glycol (PEG) moieties used in lipid nanoparticle formulations prolong circulation but can induce anti-PEG antibodies, causing accelerated blood clearance (ABC phenomenon) and hypersensitivity reactions (Irvine et al., 2020). Biodegradable lipid components and alternative stealth strategies are under active investigation to circumvent these issues (Kulkarni et al., 2018).

While chemical modifications such as pseudouridine incorporation reduce recognition by innate immune receptors, they do not fully eliminate immune activation, particularly with repeat dosing (Karikó et al., 2005b; Kaczmarek et al., 2017). This suggests that future progress requires not only molecular modifications but also systemic strategies for immune tolerance induction, including biodegradable lipid carriers and transient immunosuppression protocols. Without such measures, the durability of chronic RNA therapies will remain limited.

RNA therapeutics, particularly siRNAs and antisense oligonucleotides (ASOs), depend on Watson-Crick base pairing for specificity. However, partial complementarity can lead to unintended interactions with non-target transcripts, resulting in off-target gene silencing or aberrant splice modulation (Jackson et al., 2016; Birmingham et al., 2006). Such off-target effects can cause cytotoxicity, immune activation, and phenotypic abnormalities, complicating clinical translation. Advancements in siRNA design algorithms, incorporating seed region analysis and thermodynamic profiling, have improved specificity (Reynolds et al., 2004; Jackson and Linsley, 2010). Transcriptome-wide off-target assessment via high-throughput sequencing and bioinformatics pipelines allows empirical identification and minimization of unintended interactions (Fedorov et al., 2006; Grimm et al., 2006).

Emerging RNA editing technologies, such as CRISPR-Cas13-based platforms, promise higher precision by targeting RNA transcripts without permanent genomic alterations (Abudayyeh et al., 2017; Cox et al., 2017). Nonetheless, comprehensive off-target profiling remains imperative to ensure safety (Wang et al., 2018). Although bioinformatic algorithms have improved siRNA and ASO specificity, transcriptome-wide analyses continue to reveal low-frequency but biologically significant off-target interactions (Jackson and Linsley, 2010; Fedorov et al., 2006). A critical challenge is that even rare events may translate into long-term safety risks in chronic settings. This highlights the need for standardized preclinical off-target screening pipelines and regulatory requirements for transcriptome-level safety profiling, which remain underdeveloped.

Manufacturing RNA therapeutics at scale introduces multifaceted challenges. The core RNA synthesis step—often in vitro transcription—must maintain sequence fidelity, high purity, and controlled capping and tailing to ensure functionality and safety (Kauffman et al., 2016; Pardi et al., 2018). Impurities such as double-stranded RNA contaminants can provoke unwanted immune activation, mandating rigorous purification protocols (Karikó et al., 2005a).

Formulation with delivery vehicles, primarily lipid nanoparticles (LNPs), requires reproducible encapsulation efficiency, particle size distribution, and stability to preserve therapeutic activity and facilitate regulatory approval (Kulkarni et al., 2018; Sago et al., 2018). Microfluidic manufacturing technologies have enabled scalable, controlled LNP production with consistent quality (Valencia et al., 2012).Global manufacturing capacity constraints were exposed during the COVID-19 pandemic's mRNA vaccine rollout, underscoring the need for diversified and decentralized production infrastructure (Dolgin, 2021). Further innovation in continuous manufacturing, process analytical technologies (PAT), and supply chain robustness are essential to meet rising demand (Sharma et al., 2021).

Manufacturing challenges extend beyond cost and capacity. While LNP-based production has successfully scaled for vaccines, viral vector manufacturing remains less adaptable, constrained by packaging limits and complex purification processes (Naldini, 2015; Ginn et al., 2018). These differences underscore the importance of aligning platform choice with scalability requirements, particularly for global distribution. Without more robust and harmonized production pipelines, supply bottlenecks will continue to hinder equitable access.

Several biological and physiological barriers impede RNA therapeutic delivery to extrahepatic organs. The vascular endothelium, tissue extracellular matrix, cellular membranes, and endosomal compartments each impose restrictions on RNA biodistribution and intracellular bioavailability (Sahay et al., 2010). The liver's unique fenestrated endothelium facilitates nanoparticle and conjugate access to hepatocytes, a privilege not shared by many other tissues.

For CNS delivery, the blood-brain barrier (BBB) presents a formidable obstacle, restricting passage of most macromolecules including RNA-loaded nanoparticles. Strategies under investigation to circumvent the BBB include receptor-mediated transcytosis via transferrin or insulin receptors, intrathecal or intracerebroventricular administration, and design of ultrasmall or surface-modified nanoparticles capable of BBB penetration (Bicker et al., 2014; Saraiva et al., 2016). Nonetheless, clinical translation of CNS RNA therapeutics remains nascent, with Nusinersen—an intrathecally delivered ASO for spinal muscular atrophy—being a notable exception (Mercuri et al., 2018).

Targeting solid tumors and pulmonary tissues faces additional complexity due to heterogenous tumor vasculature, high interstitial pressure, and immune-suppressive microenvironments (Shi et al., 2020). Nanoparticle surface modifications with ligands targeting tumor-associated antigens or components of the tumor microenvironment offer promising approaches to enhance specificity and uptake (Bertrand et al., 2014). Pulmonary delivery of RNA via inhalable formulations is also being explored for respiratory diseases, but challenges in aerosol stability and mucociliary clearance remain (Lam et al., 2012).

Importantly, progress in extrahepatic delivery has been incremental rather than transformative. Most approved therapies remain liver-targeted, while CNS and solid tumor delivery are still experimental (Saraiva et al., 2016; Shi et al., 2020). This persistent gap suggests that delivery innovations should be evaluated not only for proof-of-concept efficacy but also for translational scalability, manufacturability, and safety in chronic settings—dimensions often overlooked in preclinical studies.

As RNA therapeutics transition from acute treatments to chronic disease management, long-term safety data become paramount. Chronic administration increases the risk of cumulative toxicity, immune sensitization, and unforeseen off-target effects (Setten et al., 2019). Immunogenicity, though mitigated by nucleoside modifications and delivery optimization, can evolve over prolonged exposure. Development of anti-drug antibodies or adaptive immune responses to delivery vehicle components may reduce efficacy or precipitate adverse reactions (Irvine et al., 2020). Potential genotoxicity and unintended genomic alterations are concerns particularly relevant to RNA-guided genome and epigenome editing platforms such as CRISPR-Cas9 and base editors. Although RNA editing is transient and does not alter DNA directly, delivery methods and off-target activity necessitate rigorous preclinical and clinical evaluation (Gillmore et al., 2021; Cox et al., 2017). Post-marketing surveillance programs, patient registries, and real-world data collection are critical to monitor long-term outcomes and rare adverse events, complementing pre-approval safety studies. Regulatory agencies emphasize the importance of such pharmacovigilance frameworks to ensure sustained patient safety.

Personalization is emerging as the cornerstone of next-generation RNA therapeutics. Unlike traditional small molecules or protein biologics, RNA therapies offer unparalleled modularity—enabling rapid sequence customization to match individual genetic mutations, neoantigens, or disease-specific transcriptomic profiles (Sahin and Türeci, 2018). This capability is particularly crucial in oncology, where tumor heterogeneity and patient-specific mutational burdens necessitate bespoke immunotherapies.

Personalized mRNA cancer vaccines, which encode patient-specific neoantigens identified through tumor exome sequencing, exemplify this approach. Clinical trials have demonstrated that these vaccines elicit robust and durable cytotoxic T-cell responses, often synergizing with immune checkpoint inhibitors to overcome tumor immune evasion (Ott et al., 2017; Keskin et al., 2019). The ability to rapidly synthesize and clinically deploy patient-tailored mRNA constructs is facilitated by platform manufacturing, which standardizes non-sequence elements while allowing individualized payload variation (Pardi et al., 2018).

Beyond oncology, genetic diseases with rare or private mutations benefit from personalized RNA therapies. The case of Milasen—a splice-switching antisense oligonucleotide (ASO) developed for a single patient with Batten disease—demonstrates the feasibility of ultra-rapid, patient-specific RNA therapeutic design, manufacturing, and compassionate clinical use (Kim et al., 2019). This "n-of-1" approach challenges traditional drug development paradigms, raising regulatory, ethical, and manufacturing challenges that necessitate novel frameworks for evaluation and approval (Miller et al., 2020). Realizing widespread personalized RNA therapeutics will depend on scalable, agile manufacturing technologies, streamlined regulatory pathways that accommodate rapid design cycles, and advanced biomarker-guided patient stratification (Sahin et al., 2017).

RNA editing technologies are revolutionizing the therapeutic landscape by enabling transient, reversible, and highly specific modifications of RNA transcripts without altering the underlying genomic DNA, thus offering safety advantages over permanent genome editing (Cox et al., 2017).

CRISPR-Cas13 systems, distinct from the DNA-targeting Cas9, provide programmable RNA cleavage and modulation capabilities. Cas13 effectors can be fused to RNA editing enzymes to facilitate base conversions (e.g., adenosine-to-inosine) that correct point mutations at the transcriptome level, thereby restoring protein function (Abudayyeh et al., 2017; Konermann et al., 2018). These RNA-targeting systems minimize risks of off-target DNA damage and heritable changes, offering a therapeutic advantage for transient or dosage-dependent disease modulation (Liu et al., 2017).

Moreover, base editors such as REPAIR and RESCUE enable single-nucleotide modifications with high specificity, broadening the therapeutic window for diseases caused by point mutations (Zhang et al., 2017). Delivery of these complex editing constructs remains a critical bottleneck; lipid nanoparticle encapsulation and adeno-associated viral vectors are being optimized for tissue-specific delivery and immune evasion (Gillmore et al., 2021). Ethical and safety considerations surrounding RNA editing include immunogenicity of CRISPR components, off-target RNA binding, and unintended splice variants. Ongoing preclinical and clinical studies aim to characterize and mitigate these risks (Liang et al., 2017). Beyond editing systems, additional RNA modalities such as circRNAs, lncRNAs, riboswitches, and RNA-targeting small molecules are expanding the therapeutic landscape; each bringing distinct opportunities and challenges for translation.

Taken together, these challenges are unevenly distributed across RNA modalities. Immunogenicity is a well-characterized concern for mRNA and saRNA vaccines, while CRISPR-based editors remain in earlier stages of evaluation where long-term safety and immune responses are less defined. Manufacturing scalability is largely solved for mRNA/LNPs, remains cost-intensive for AAV, and is still immature for exosome-based systems. Delivery beyond the liver continues to limit all platforms, with GalNAc confined to hepatocytes and polymers or exosomes showing only exploratory promise. Long-term safety data are strongest for ASOs and siRNAs, whereas newer platforms such as RNA editors and circRNAs lack longitudinal follow-up. These cross-modal comparisons are consolidated in Table 3, which links each challenge to the modalities most affected and the maturity of the supporting evidence.

Artificial intelligence (AI) and machine learning (ML) technologies are fundamentally transforming RNA therapeutic development by enabling data-driven design, predictive modeling, and personalized treatment strategies. AI algorithms facilitate rational design of RNA sequences that maximize target affinity while minimizing immunogenicity and off-target effects, leveraging deep learning models trained on extensive datasets of RNA structure, function, and biological interactions. These computational tools reduce iterative laboratory experimentation, accelerating candidate optimization and reducing development timelines.

Recent studies highlight the tangible benefits of AI integration in RNA therapeutic development. For example, deep learning–based siRNA design platforms have reduced predicted off-target binding by up to 40%–60% compared with traditional rule-based algorithms. Similarly, generative AI models have been applied to optimize mRNA secondary structure, resulting in twofold improvements in protein translation efficiency in preclinical assays. Machine learning frameworks have also predicted LNP biodistribution patterns with >80% accuracy, accelerating formulation development and reducing empirical trial-and-error. These case studies illustrate that AI not only accelerates discovery but also quantitatively improves therapeutic design and delivery performance.

In delivery system development, ML models predict nanoparticle physicochemical properties, biodistribution patterns, and cellular uptake efficiency, enabling in silico screening of formulation parameters. This reduces reliance on costly and time-intensive empirical trials. Clinically, AI supports patient stratification through integration of multi-omics data, electronic health records, and imaging, identifying responders to RNA therapeutics and predicting adverse events (Topol, 2019). Adaptive trial designs powered by AI enhance statistical power and personalized dosing regimens, improving therapeutic outcomes and reducing trial failures. The future convergence of RNA therapeutics with AI-driven precision medicine heralds an era where RNA drugs are designed, delivered, and dosed with unparalleled specificity, heralding a new paradigm of individualized healthcare.