mRNA vaccines proved RNA-based medicines could work at scale. But RNA therapeutics have moved well beyond that single modality, with siRNA and antisense oligonucleotides (ASOs) now carrying multiple approved products and pipelines expanding into cardiometabolic, central nervous system (CNS), and rare liver diseases.
What has changed isn't just the science, according to Jing Zhu, vice president of technical operations and commercial strategy at Hongene Biotech, but manufacturability. "These molecules aren't just scientifically interesting anymore, but they're actually manufacturable at scale, which matters more than people give it credit for," Zhu told DDN, pointing to GalNAc conjugation as a case in point: the approach simplified liver-targeted delivery relative to lipid nanoparticles, streamlining the path from chemistry to manufacturing.
Beyond siRNA and ASOs, newer modalities including circular RNA, self-amplifying RNA, small activating RNAs, and aptamers are gaining traction, each with distinct manufacturing challenges. Self-amplifying constructs, for example, can be dosed at lower levels, but construct design and manufacturing scale-up become more difficult.
Delivery beyond the liver remains unsolved
Hepatic delivery has advanced considerably thanks to lipid nanoparticles and GalNAc conjugates, but delivering RNA therapeutics reliably to muscle, the CNS, or the lungs remains a largely unsolved problem, Zhu said. Manufacturing experience in those areas is still limited, raw materials less standardized, and analytical methods still maturing.
Stability presents a related challenge. RNA is vulnerable to nucleases and hydrolysis, which puts pressure on formulation, cold chain, and shelf life. Chemical modifications used to address that instability, such as 2'-O-methyl, 2'-fluoro, and phosphorothioate backbone modifications, each add complexity that must hold up at commercial scale rather than only in small batches.
The field is now designing modifications and delivery vehicles with manufacturability built in from the outset, rather than optimizing for potency alone and addressing production later, Zhu said. That is paired with investment in raw material quality, characterization tools, and process intensification aimed at reducing batch-to-batch variability.
Chemistry advances are reshaping manufacturing and QC
Expanded chemical modification options have made RNA molecules more nuclease-resistant, less immunogenic, and easier to fine-tune pharmacokinetically, opening up targets that were previously out of reach. But those same modifications complicate synthesis, introduce more impurities to track, and raise the bar for purification and analytical work.
"Chemistries that actually make it to the clinic are increasingly the ones designed with process feasibility in mind from the start," Zhu said, noting that a modification that performs well in a test tube but proves difficult or costly to manufacture is unlikely to advance. Improvements in solid-phase synthesis, enzymatic and chemical ligation methods, and more sensitive analytical platforms have helped heavily modified RNA constructs move from academic research to GMP-manufacturable clinical candidates on more realistic timelines.
AI is compressing the path from target to candidate
The process of moving from RNA target identification to clinical candidate looks different than it did five years ago, according to Zhu. Teams previously optimized for potency and specificity early, only to encounter manufacturing setbacks at scale-up that triggered costly late-stage redesigns.
Machine learning models are now used earlier in that process to predict secondary structure, off-target effects, and modification patterns before synthesis begins, allowing teams to screen more sequence and chemistry variants virtually before committing lab resources. Manufacturability and analytical characterization are increasingly folded in at the lead selection stage, with AI-assisted modeling flagging potential process or purification issues earlier. Maturing supply chains for RNA building blocks have also shortened early feasibility timelines that were previously a monthslong bottleneck.
What it means for drug developers
Looking ahead, Zhu pointed to a shift beyond rare disease into higher-prevalence indications such as cardiovascular disease, chronic liver conditions, and select oncology and infectious disease programs. That shift changes the manufacturing calculus: rare disease programs can accommodate smaller-scale, higher-cost production given small patient populations, but common disease indications require processes and supply chains built for commercial scale at a cost structure that supports broad access.
AI is expected to extend further into manufacturing itself, Zhu said, including predictive process modeling to flag yield or purity issues before a batch runs, digital twins of production processes for virtual testing, and real-time analytics that catch quality drift during production rather than after. Combined with continuous manufacturing and more automated processing, Zhu said the next several years are likely to focus less on new chemistry breakthroughs alone and more on industrializing RNA therapeutics for mainstream, high-volume diseases.
