Why RNA–LNP Development Needs Higher-Resolution Analytics In 2027

By Partha Anbil, Life Sciences Industry Advisor, MIT Sloan Career Development Office

RNA is a modality that has outgrown its measurement tools.

RNA has quietly become one of the most productive areas in drug development. Twenty-five RNA therapeutics now hold FDA approval, roughly 734 trials are running worldwide, and 2025 delivered a record of about 164 new trial starts. Sales excluding prophylactic vaccines passed $7 billion in 2025, while RNA licensing deal value exceeded $17 billion, and emerging biopharma — not large pharma — sponsors more than 55% of active trials.¹

The center of gravity has shifted with the volume. siRNA accounts for 37% of ongoing trials and antisense oligonucleotides another 33%, while mRNA sits at 13%, with circRNA, miRNA, and RNA editing entering the clinic behind them.¹ Indications have moved out of rare disease into cardiovascular, metabolic, oncology, and CNS territory — patient populations measured in millions rather than thousands, at doses and durations that leave far less room for a formulation that is merely adequate.

Figure 1: Modality mix of ongoing RNA clinical trials, 2026. Oligonucleotide modalities account for roughly seven in 10 active trials. Data source: IQVIA Institute, March 2026.¹

That commercial position rests on a delivery technology whose quality attributes are still measured coarsely. Most RNA drugs that reach the cytoplasm at scale do so inside a lipid nanoparticle (LNP),² and the LNP is precisely where drug substance and drug product stop being separable problems. Yet in most early development laboratories the analytical package for an LNP remains three ensemble measurements: a z-average diameter and polydispersity index (PDI) by dynamic light scattering (DLS), a zeta potential by electrophoretic light scattering, and an encapsulation efficiency by fluorescent dye exclusion. Those three numbers were adequate when the LNP was a vaccine-era convenience. They are not adequate now that the LNP is the product.

The Averaging Problem

Each of those routine assays reports a population average, and each hides something different.

DLS returns an intensity-weighted mean. Because scattering intensity scales with roughly the sixth power of diameter, a small number of large particles dominates the signal. Two formulations with indistinguishable z-averages and PDI values below 0.1 can have materially different subpopulation structures, and PDI itself is a fitted width, not a distribution — it cannot say whether a sample is one narrow population or two overlapping ones.

Dye-exclusion encapsulation assays have a subtler failure mode. They report the proportion of recovered RNA protected from the dye, not the proportion of charged RNA that survived formulation, purification, and filtration. It is entirely possible — and in practice common — to report encapsulation efficiency above 97% alongside a mass balance recovery near 40%. Both numbers are correct. Read together, they say that most of the RNA put into the process never reached the vial. Read apart, the first says the formulation is excellent.

Zeta potential shares the same character: an ensemble electrophoretic mobility over a heterogeneous population, blind to whether surface charge varies between subpopulations.

Nanoparticle tracking analysis is a genuine step up, because it sizes particles one at a time from their diffusion and returns a number-weighted distribution and a particle concentration that DLS cannot provide. But it inherits its own artifacts. At the concentrations LNP samples usually require, coincident scattering events merge neighboring particles into apparent larger ones; below about 50 nm, detection efficiency falls away; and the dilution needed to reach a workable field of view can itself perturb a colloidally marginal formulation.³ NTA sharpens the picture of size. It still says nothing about morphology, internal architecture, or how much RNA any individual particle is carrying.

None of this makes the routine assays wrong. They are fast, inexpensive, transferable, and well suited to release testing against an established specification. The problem is using release-grade resolution to make discovery-grade decisions — ranking candidates, locking a process, judging comparability after a scale change — where the differences that matter are distributional rather than average.

Heterogeneity Is A Potency Variable, Not A Curiosity

Recent single-particle work makes the cost of averaging concrete. Counting particles one at a time by cylindrical illumination confocal spectroscopy, a 2026 study found that at a 1 mL/min microfluidic mixing rate, roughly 83% of LNPs were empty, and that raising the total flow rate to 30 mL/min sharply reduced that empty fraction — with no meaningful change in particle size.⁴ The attribute that moved was invisible to the assay a QC laboratory would have used to judge the change.

The functional consequence appeared only under stress. At a saturating siRNA dose, every formulation produced equivalent knockdown; at sub-saturating doses, particles made at the higher flow rate consistently outperformed the rest.⁴ A potency assay run at a comfortable dose would have called the formulations equivalent. So would DLS. Independent work published in Molecular Therapy in 2026 reaches a parallel conclusion for gene editing: how mRNA and guide RNA are distributed across particles — not how much total RNA is present — drives editing efficiency in vivo.⁵

What makes this actionable is that the levers are ordinary process parameters already in every formulation scientist's hands. Choice of ionizable lipid, nitrogen-to-phosphate ratio, aqueous-to-organic flow rate ratio, and microfluidic total flow rate all move particle size, size distribution, and morphology, and they do not move them together. Reducing total flow rate can add 10 nanometers to median diameter while leaving PDI, zeta potential, and encapsulation efficiency essentially unchanged — a shift a design-of-experiments screen built on those three readouts would score as no effect at all. Downstream purification, buffer exchange, and cryoprotectant level exert comparable influence on what survives to the vial.

The implication for development strategy is direct. Payload distribution behaves like a critical quality attribute (CQA). It responds to process parameters, it varies between batches, and it changes efficacy at clinically relevant doses. If it is not measured it is not controlled, and it resurfaces later as unexplained batch-to-batch potency variation, a failed comparability exercise after tech transfer, or a dose response curve that shifts between pilot and commercial scale.

What Higher Resolution Actually Looks Like

The answer is not to replace the routine assays but to layer an orthogonal, higher-resolution tier above them at the points where a decision is expensive to reverse.

On the drug substance side, hydrophilic interaction chromatography coupled to mass spectrometry (HILIC-MS) resolves oligonucleotide impurities — n-1 shortmers, depurination products, incomplete sulfurization — that ion-pairing reversed-phase UV alone reports as a single peak envelope.⁶ High-resolution Orbitrap platforms give the most confident assignments, but the accessibility gap is real: cost and specialist expertise keep them out of many formulation laboratories. Comparative work indicates that properly optimized single-quadrupole systems can carry a meaningful share of the routine impurity workload⁷ — which matters more to the emerging biopharma sponsors running most of the pipeline than another increment of mass accuracy.

Figure 2: Critical quality attribute coverage by analytical method. No single technique spans the attribute set; the case for an orthogonal pipeline is the pattern of gaps, not the merit of any one row.

On the drug product side, field-flow fractionation is the pivotal technique. Asymmetric-flow FFF (AF4) physically separates particles by hydrodynamic size before detection, removing the coincident-scattering artifacts that limit DLS and nanoparticle tracking analysis. Coupled online to UV, multi-angle light scattering (MALS), and DLS, it yields a fractionated size distribution, a radius of gyration, and a hydrodynamic radius across the elution profile and, critically, the ratio between them — a shape factor that separates a dense core–shell particle from a spherical one of the same nominal diameter. Dual-detector UV/MALS signals support estimates of RNA weight fraction and RNA copies per particle across the distribution, information no bulk encapsulation assay produces.³ Electrical AF4 adds population-resolved charge. This is not exotic instrumentation: AF4 is codified in ISO 21362⁸ and has been assessed jointly by the EU Nanomedicine Characterization Laboratory and the U.S. National Cancer Institute's Nanotechnology Characterization Laboratory as a method for particle size, drug loading, and stability of nanopharmaceuticals.⁹

Two practical cautions apply. AF4 is a separation, and separations can lose sample: cationic formulations interact electrostatically with the anionic separation membrane, and recovery must be established, and often engineered through membrane preconditioning, before any distribution is believed. Second, no single method closes the question. Cryogenic electron microscopy remains the reference for internal architecture and small-angle X-ray scattering for lipid packing; both are low-throughput and belong at milestones, not in a screen. The point of an orthogonal pipeline is agreement between independent physical principles, not a single higher number.

Table 1: The RNA–LNP analytical toolkit: what each tier actually delivers.

The Regulatory Pull Is Now Real

Three developments have turned analytical resolution from a scientific preference into a filing consideration.

ICH Q14 and the revised Q2(R2) move analytical procedures onto a life cycle footing, asking sponsors to justify a method against the attribute it controls and the range over which it must perform.¹⁰ An enhanced-approach submission that claims control of particle size distribution on the strength of a single z-average invites an obvious question about what else that distribution contains.

The EMA's draft guideline on the quality aspects of mRNA vaccines, issued in March 2025 with consultation closing that September and revisions expected before finalization, tightens expectations on raw material testing, in-process monitoring, batch consistency, and the evidence needed to justify a platform claim.¹¹ Commentary through early 2026 reads it as the end of the pandemic-era accelerated posture and the beginning of a longer-term compliance model — one that falls proportionally harder on smaller sponsors.¹² USP's Analytical Procedures for the Quality of mRNA Vaccines and Therapeutics, now in a third draft edition, is building the method-level scaffolding underneath it.¹³

The funding environment sharpens the point. HHS canceled roughly $500 million across 22 BARDA-funded mRNA vaccine projects in August 2025,¹⁴ and capital has rotated toward siRNA, ASO, and non-vaccine mRNA applications. In a market where more than half of trials are run by emerging companies competing for that capital, early high-resolution evidence that a candidate is well characterized and its process understood is a commercial asset, not an overhead line.

Table 2: What the 2026 regulatory and standards environment asks of RNA–LNP analytics.

5 Moves Worth Making Now

• Separate release analytics from decision analytics. Keep DLS, ELS, and dye-exclusion assays for what they do well. Add a fractionating orthogonal method at the three points where a wrong call is costly: candidate selection, process lock, and post-scale-up comparability.
• Never report encapsulation efficiency without mass balance. The pair is interpretable; either number alone is not. Make both a standing output of the formulation record, not a number retrieved on request.
• Qualify one high-resolution method early, then anchor it. Build a correlation between AF4-derived distribution parameters and the routine QC readouts on your own formulations. That correlation lets a fast assay stand in for a slow one later, and it is close to the argument a Q14 submission wants to see.
• Test whether raw-material variability actually propagates. Vendor-to-vendor differences in RNA drug substance chain length and molecular weight can be substantial and still yield indistinguishable drug product under controlled microfluidic manufacture. Characterize inputs thoroughly, but qualify vendors on drug product attributes rather than drug substance specifications alone.
• Start accumulating the data set. Predictive models of formulation performance are the obvious destination, and the limiting factor is not algorithms but structured CQA data with consistent provenance across formulations, processes, and stability conditions. Organizations that begin building that library now will be able to model in three years; those that do not will still be running one-off screens.

The Bottom Line

RNA–LNP therapeutics are complex products characterized, in most laboratories, with tools built for simpler ones. That mismatch is not a laboratory inconvenience; it is a translation risk carried on the balance sheet, because the decisions it corrupts — which candidate advances, which process is locked, whether two batches are the same — are the expensive ones. The gap does not announce itself as a failed release test. It appears later — as attrition, as an unexplained potency shift, as a comparability exercise that will not close. Raising analytical resolution at the point of decision is among the cheapest interventions available, and in 2026, with regulators tightening, capital selective, and the modality moving into large populations, it is also among the most consequential.

References

1. IQVIA. The Next Frontier of RNA Therapeutics. IQVIA Institute blog, March 2026. https://www.iqvia.com/locations/emea/blogs/2026/03/the-next-frontier-of-rna-therapeutics
2. Kulkarni JA, Witzigmann D, Thomson SB, Chen S, Leavitt BR, Cullis PR, et al. The current landscape of nucleic acid therapeutics. Nature Nanotechnology. 2021;16(6):630–43. doi:10.1038/s41565-021-00898-0
3. Parot J, et al. Improved multidetector asymmetrical-flow field-flow fractionation method for particle sizing and concentration measurements of lipid-based nanocarriers for RNA delivery. European Journal of Pharmaceutics and Biopharmaceutics. 2021;163:252–265.
4. Pial, Li, Lin, Wang, Mao, Curk, et al. Controlling Payload Heterogeneity in Lipid Nanoparticles for RNA-Based Therapeutics. Advanced Functional Materials. 2026. doi:10.1002/adfm.202526278 (preprint: bioRxiv 2025.06.11.659145)
5. Messenger RNA and guide RNA distributions in lipid nanoparticles impact gene-editing efficiency in vivo. Molecular Therapy. 2026. https://www.cell.com/molecular-therapy-family/molecular-therapy/fulltext/S1525-0016(26)00293-5
6. Advances in Analysis of Therapeutic Oligonucleotides with Chromatography Coupled to Mass Spectrometry. Analytical Chemistry. 2026;98(15):10895. https://pubs.acs.org/ancham/article/98/15/10895/5136644/Advances-in-Analysis-of-Therapeutic
7. Camperi J, Lippold S, Ayalew L, Roper B, Shao S, Freund E, et al. Comprehensive Impurity Profiling of mRNA: Evaluating Current Technologies and Advanced Analytical Techniques. Analytical Chemistry. 2024;96(9):3886–97. doi:10.1021/acs.analchem.3c05539
8. ISO 21362:2018. Nanotechnologies — Analysis of nano-objects using asymmetrical-flow and centrifugal field-flow fractionation. International Organization for Standardization.
9. Asymmetric-flow field-flow fractionation for measuring particle size, drug loading and (in)stability of nanopharmaceuticals: the joint view of the European Union Nanomedicine Characterization Laboratory and the National Cancer Institute Nanotechnology Characterization Laboratory. Journal of Chromatography A. 2021. https://www.sciencedirect.com/science/article/pii/S0021967320310414
10. International Council for Harmonization. ICH Q14: Analytical Procedure Development, and ICH Q2(R2): Validation of Analytical Procedures. https://database.ich.org
11. European Medicines Agency. Draft guideline on the quality aspects of mRNA vaccines. EMA/CHMP, March 2025. https://www.ema.europa.eu/en/documents/scientific-guideline/draft-guideline-quality-aspects-mrna-vaccines_en.pdf
12. mRNA makers face a more exacting Europe. The mRNA Conference Europe, 12 February 2026. https://www.europe.the-mrna-conference.com/news/mrna-makers-face-a-more-exacting-europe
13. United States Pharmacopeia. Analytical Procedures for the Quality of mRNA Vaccines and Therapeutics (Draft Guidelines, 3rd Edition). USP-NF notice, 2 August 2024. https://www.uspnf.com/notices/analytical-procedures-mrna-vaccines-20240802
14. HHS Announces Reduction in mRNA Vaccine Development Programs Under BARDA, Nearly $500 Million in Contracts Canceled. Pharmaceutical Executive, August 2025. https://www.pharmexec.com/view/hhs-reduction-mrna-vaccine-development-barda-500-million-contracts-cancelled

Disclaimer: The views expressed in the article are those of the author and not of the organizations they represent.

About The Author:

Partha Anbil is at the intersection of the life sciences industry and management consulting, with over 30 years of experience in life sciences. He is also a life sciences industry advisor at MIT, his alma mater. He held senior leadership roles at WNS, IBM, Booz & Company, Symphony, IQVIA, KPMG Consulting, and PWC. Anbil has consulted with and counseled health and life sciences clients on structuring solutions to address strategic, operational, and organizational challenges. He is a diplomat-in-residence and fellow at MIT CSAIL and a healthcare expert member of the World Economic Forum (WEF). He was a member of the IBM Industry Academy, a highly selective group of professionals inducted by invitation only and considered IBM's highest honor.