The expression window as a safety parameter for mRNA medicines.

Messenger RNA therapeutics are engineered, with few exceptions, for stability and output. For an expanding class of payloads that objective is wrong. Once a nuclease has finished editing, a cytokine has initiated a response, or an in vivo generated receptor has cleared its target, continued expression adds no benefit while the liability accumulates. The goal is then predictable disappearance, and the quantity to engineer is the expression window: the time course of active protein. This review formulates it as the convolution of three layers. Formulation and route supply the input function, dominate the rise and peak, and can dominate the apparent decay; once release is complete, RNA and protein turnover set the tail, which declines at the slower of the two rates. Levers at these layers are not substitutes. In a protein-limited window, RNA destabilisation leaves the terminal decay rate unchanged and reduces exposure almost entirely by lowering the peak. In the worked example, each unit of peak surrendered buys about a sixth of the tail reduction available at the protein layer, and the ranking reverses when the RNA is limiting. Choosing a lever therefore requires knowing which layer is limiting, which a paired RNA-and-protein time course decides once release is separated from decay. Three reportable metrics - rise, peak and offset time - replace the single stated duration, and cumulative exposure, not peak, matches an accumulating liability. The asymmetry is structural: of the established RNA-layer levers reviewed here, every one extends the window or restricts it in space.