How Should We Measure The True Potency Of RNA Therapeutics?

By Jyotsna Jajula, research assistant, Wayne State University

An RNA therapeutic can generate a strong molecular signal without producing the intended biological function. An mRNA may yield abundant protein that lacks the necessary activity, an siRNA may reduce its target transcript without sufficiently depleting the corresponding protein, and a splice-modulating oligonucleotide may correct RNA processing without restoring a functional phenotype. Expression or target engagement demonstrates molecular activity, but neither alone establishes potency.

This distinction is becoming more consequential as RNA medicines expand into protein replacement, gene silencing, personalized cancer immunotherapy, transient cell programming, and in vivo genome editing. Each modality creates a different relationship between molecular output and therapeutic function. A cancer vaccine must convert antigen expression into immune recognition. A protein replacement therapy must produce correctly processed protein at a functionally sufficient level. An RNA interference therapy must translate transcript reduction into protein depletion and pathway modulation.

Potency should therefore describe a product’s capacity to produce its intended biological activity at a defined dose under controlled conditions. It should remain distinct from clinical efficacy, but the selected measurement must occupy a defensible position in the product’s mechanism of action. The central challenge is not simply choosing a more complex assay. It is determining which features of the response — magnitude, functional threshold, responder fraction, and duration — best predict whether the RNA has completed its assigned biological task.

A High Signal Does Not Necessarily Mean High Potency

Expression measurements often rank candidates according to total molecular output. This approach assumes that a larger signal represents a proportionally stronger product. Biological systems rarely behave so simply. The relationship between RNA dose, molecular output, and functional response may contain thresholds, plateaus, delays, or feedback mechanisms that alter how potency should be interpreted.

A threshold response occurs when molecular activity must exceed a minimum level before a meaningful phenotype appears. An enzyme replacement strategy may require sufficient active protein to restore pathway flux, while lower expression remains biologically inconsequential. A plateau occurs when the relevant pathway becomes saturated: additional protein expression increases the analytical signal without improving function. Excessive activity may even become undesirable if it disrupts pathway balance or intensifies innate immune signaling.

A single high-dose measurement cannot reveal these relationships. Developers should compare candidates across a dose range and characterize both the amount of product required to generate a response and the maximum functional response the product can achieve. Two constructs may reach the same peak expression but differ in the dose needed to cross the functional threshold. Conversely, a candidate with lower maximum expression may generate greater activity per unit dose because the encoded protein is processed or functions more efficiently.

This distinction is particularly important when sequence engineering changes expression kinetics. Coding sequence structure and untranslated regions can influence translational efficiency and functional mRNA half-life, altering total protein output without changing the encoded protein itself. Mauger and colleagues demonstrated that mRNA structure can affect protein expression through changes in functional half-life.¹ Such findings show why measuring protein at one time point cannot fully characterize the biological behavior of an mRNA construct.

A scientifically informative potency profile should therefore distinguish at least three properties: the dose required to initiate function, the maximum functional response, and the range over which response changes with dose. These properties can influence candidate selection even when conventional expression measurements appear similar.

Potency Has A Distribution, Not Just An Average

Bulk measurements compress heterogeneous cellular responses into a single value. A twofold increase in mean protein expression can result from moderate activity across most cells or intense expression in a small responder population. Likewise, 70% target knockdown in a mixed cell population may indicate partial silencing in nearly every cell or near-complete silencing in one subset. These patterns are numerically similar at the population level but may not be biologically equivalent.

The preferred distribution depends on the therapeutic task. A limited population of highly productive cells may supply sufficient quantities of a secreted protein. Correction of a cell-autonomous defect may instead require activity across a substantial fraction of affected cells. Cancer immunotherapy may depend on which antigen-presenting or immune cell populations respond rather than on the largest average signal across all cells.

Responder frequency and per-cell response intensity should therefore be treated as separate potency dimensions. Multiparameter flow cytometry can determine whether the relevant cellular population responds and quantify the distribution of activity within that population. Quantitative imaging can retain spatial information that disappears during tissue homogenization. Single-cell approaches can identify rare high-output populations that disproportionately influence a bulk measurement.

This resolution can change development decisions. A formulation that produces a higher average signal through extreme activity in a small subset may be less suitable than one that generates moderate but consistent activity across the required cells. Alternatively, a secreted protein therapy may benefit from concentrated activity if those responder cells can sustain adequate systemic protein levels. The correct interpretation follows from the mechanism, not from a universal preference for either breadth or intensity.

Developers should consequently define two thresholds when the biology supports them: the minimum per-cell activity required for function and the minimum fraction of relevant cells that must cross that threshold. This approach transforms cellular heterogeneity from an uncontrolled source of variability into a measurable product characteristic.

The Time Course Can Reverse A Candidate Ranking

Potency is also a temporal property. A candidate that produces the highest peak may not provide the most useful activity over the period required for treatment. Ranking constructs at a single time point can therefore favor a transient high-output candidate over one that produces a lower but more appropriate functional trajectory.

The relevant kinetic profile varies by modality. Protein replacement may require activity to remain above a minimum threshold between doses. RNA interference must maintain protein depletion long enough to alter the targeted pathway, particularly when the existing protein turns over slowly. Vaccination requires sufficient antigen production to support immune priming, but persistent inflammatory signaling may offer no additional benefit. Self-amplifying RNA further complicates interpretation because intracellular amplification can change the relationship among administered dose, antigen production kinetics, and immune response. A Phase 3 comparison of the self-amplifying RNA vaccine ARCT-154 and BNT162b2 illustrates why neutralizing antibody responses provide more meaningful functional evidence than intracellular RNA amplification alone.²

Transient expression can be a design objective rather than a limitation. Rurik and colleagues used modified mRNA to generate chimeric antigen receptor T cells in vivo in a preclinical model of cardiac injury. The temporary expression program produced a cellular effect without permanently modifying the T cells.³ Here, potency cannot be reduced to the duration of chimeric antigen receptor expression; the relevant question is whether transient programming generates enough functional cells for long enough to modify the pathological process.

In genome editing, a short exposure can create a durable outcome. Clinical evaluation of an in vivo CRISPR therapy for transthyretin amyloidosis used transient Cas9 mRNA and guide RNA delivery, yet the resulting genomic modification produced sustained reduction of serum transthyretin.⁴ Continued editor expression was unnecessary because the biological consequence persisted after the RNA had disappeared.

Potency studies should therefore capture a response trajectory rather than a convenient peak. Useful parameters include onset, maximum activity, time above the functional threshold, cumulative functional output, and return toward baseline. The optimal trajectory is not necessarily the longest. It is the trajectory that matches the mechanism while avoiding unnecessary exposure.

The Functional Endpoint Must Match The Modality

No universal endpoint can define potency across RNA medicines because different modalities perform different biological tasks. The strongest strategy begins by identifying the final measurable function that remains sufficiently close to the product’s mechanism of action.

For protein replacement, total protein concentration may be inadequate if folding, processing, localization, or enzymatic activity determines function. A relevant potency measurement might quantify substrate conversion, receptor activation, pathway restoration, or secretion of active protein. For RNA interference, teams should connect transcript reduction to protein depletion and, when feasible, a downstream phenotype. The interval between these measurements matters because mRNA and protein can decay at different rates.

For splice modulation, the percentage of corrected transcript should be related to restored protein and cellular function. For genome editing, editor expression is an intermediate event; on-target editing in the relevant cell population and the resulting functional correction provide stronger potency evidence. Off-target editing belongs primarily to specificity and safety assessment and should not be misclassified as potency.

Cancer vaccines require a different hierarchy. Antigen expression confirms translation, but functional potency depends on antigen processing, presentation, and activation of relevant immune responses. The Phase 2b KEYNOTE-942 study of individualized mRNA-4157/V940 combined with pembrolizumab illustrates the translational importance of this hierarchy: the therapeutic premise was not simply production of patient-specific neoantigens but generation of antitumor immunity associated with improved recurrence-free survival.⁵ Clinical outcome does not replace a potency assay, but it clarifies the biological function that preclinical and analytical measurements should represent.

The most informative endpoint is therefore not always the closest measurement to the RNA molecule. It is the endpoint that captures the product’s intended activity without becoming so distant that unrelated biological variables dominate the result.

Build A Decision-Grade Potency Framework

Development teams can define a potency strategy through five linked questions:

• What immediate molecular event does the RNA produce?
• What functional event must follow?
• What dose–response relationship connects those events?
• How many relevant cells must respond, and to what extent?
• How long must the functional response remain above its required threshold?

These questions create a therapeutic task map that can guide candidate ranking and later analytical development. Early studies may use several orthogonal measurements to identify which mechanistic step best differentiates candidates. For example, an mRNA program could combine protein quantity, active protein function, responder frequency, and response kinetics. These measurements should not be collapsed prematurely into a single value; each reveals a different reason why a candidate may succeed or fail.

As the program matures, developers can select a simpler primary potency readout if they demonstrate that it remains connected to functional activity. More complex assays can retain a supporting role by confirming that the relationship remains valid after formulation or manufacturing changes. This tiered approach avoids requiring one assay to reproduce the entire therapeutic mechanism while preventing convenient molecular measurements from becoming disconnected from biological function.

A reference standard can further improve interpretation by showing whether a test article produces a comparable dose–response relationship under controlled conditions. Relative potency, however, remains meaningful only when the reference and test samples generate parallel and biologically relevant responses. A precise numerical result cannot compensate for an endpoint that poorly represents the mechanism.

The practical development question should therefore change from “Which candidate generates the largest signal?” to “Which candidate reaches the required functional threshold, in enough relevant cells, for the necessary period, at an acceptable dose?” That question is more demanding, but it produces evidence that is more useful for candidate selection, comparability, and translation.

Beyond Expression

Expression, transcript knockdown, splice correction, and editor production remain valuable measurements, but they describe the beginning of biological activity rather than its complete meaning. True potency depends on the quantitative relationship between dose and function, the distribution of that function across relevant cells, and the time for which it remains biologically useful.

As RNA medicines become increasingly programmable, potency strategies must become equally precise. The most useful measurement is not necessarily the largest signal or the most sophisticated assay. It is the measurement that most convincingly demonstrates that the RNA product has completed the biological task for which it was designed.

References:

1. Mauger DM, et al. mRNA structure regulates protein expression through changes in functional half-life. Proceedings of the National Academy of Sciences. 2019;116(48):24075–24083. doi:10.1073/pnas.1908052116.
2. Oda Y, et al. Immunogenicity and safety of a self-amplifying RNA COVID-19 vaccine (ARCT-154) versus BNT162b2: a double-blind, multicentre, randomised, controlled, phase 3, non-inferiority trial. The Lancet Infectious Diseases. 2024;24(4):351–360. doi:10.1016/S1473-3099(23)00650-3.
3. Rurik JG, et al. CAR T cells produced in vivo to treat cardiac injury. Science. 2022;375(6576):91–96. doi:10.1126/science.abm0594.
4. Gillmore JD, et al. CRISPR-Cas9 in vivo gene editing for transthyretin amyloidosis. The New England Journal of Medicine. 2021;385(6):493–502. doi:10.1056/NEJMoa2107454.
5. Weber JS, et al. Individualised neoantigen therapy mRNA-4157 (V940) plus pembrolizumab versus pembrolizumab monotherapy in resected melanoma (KEYNOTE-942): a randomised, phase 2b study. The Lancet. 2024;403(10427):632–644. doi:10.1016/S0140-6736(23)02268-7.

About The Author

Jyotsna Jajula is a research assistant at Wayne State University. Her work broadly explores RNA delivery mechanisms in oncology cell models, with a focus on internalization and cytoplasmic fate of therapeutic peptides. Jajula holds a master’s degree in pharmaceutical sciences and has prior research experience in lipid nanoparticles, RNA stability, and biodistribution strategies across oncology, immunology, and gene-therapy applications.