siRNA development

The mechanism, and the two problems it creates

A small interfering RNA directs the cell's own machinery to degrade a chosen messenger RNA, reducing production of the protein it encodes. Because recognition is by base pairing rather than by shape, a target that offers no binding pocket for a small molecule and no accessible surface for an antibody is still addressable. Two problems follow from the same property. Base pairing that is close but not exact can silence the wrong transcript, and a charged oligonucleotide does not enter cells on its own.

The approved products, read as a technical progression

Patisiran

Patisiran was the first siRNA therapeutic the FDA approved, indicated for the polyneuropathy of hereditary transthyretin-mediated amyloidosis in adults.^[S1] Its label describes a double-stranded siRNA formulated as a lipid complex for delivery to hepatocytes.^[S2] The formulation is administered by infusion, which is a consequence of the delivery vehicle rather than of the mechanism.

Givosiran

Givosiran is indicated for adults with acute hepatic porphyria.^[S3] Its label describes it as an aminolevulinate synthase 1-directed siRNA covalently linked to a ligand containing three N-acetylgalactosamine residues, to enable delivery of the siRNA.^[S4] That ligand is recognized by the asialoglycoprotein receptor on hepatocytes, an approach reviewed as one widely used strategy for delivering therapeutic nucleic acids to that cell type.^[S8] No particle is involved: the targeting is chemistry attached to the molecule itself.

Vutrisiran

Vutrisiran is a transthyretin-directed siRNA indicated for the polyneuropathy of hereditary transthyretin-mediated amyloidosis in adults.^[S5] Its label describes a chemically modified double-stranded siRNA covalently linked to a ligand containing three N-acetylgalactosamine residues.^[S6] It addresses the same target as patisiran through a different delivery approach and with a longer dosing interval. This article records that as a factual difference between the two labels and makes no claim about which is preferable.

Two things this sequence of products shows

First, a technical transition is visible directly in the labels. The earliest of the three is formulated as a lipid complex;^[S2] the two later ones are conjugates carrying a sugar ligand and no particle at all.^[S4][S6] A conjugate and a particle are different products from a nonclinical standpoint: the particle has its own distribution and its own toxicity, the conjugate does not, and the analytical methods differ throughout. Second, all three were developed by the same company. That is a fact about where practical experience in this modality is concentrated rather than a comment on any product, and it is worth knowing when assessing how much independent precedent exists for a given approach.

The problems specific to this modality

Delivery determines which organs are reachable

An unmodified siRNA is taken up poorly and degraded quickly, so delivery is a design problem that has to be solved before the sequence matters. Both approved routes shown above deliver to hepatocytes. That is the honest state of the field as reflected in the approved products: the liver is reachable, and other organs are the subject of ongoing work rather than settled practice. A programme aimed at an extrahepatic target is therefore taking on a delivery problem as well as a target problem, and the nonclinical plan should reflect that both are unresolved.

Off-target silencing

Off-target silencing is treated in the literature as one of the defining performance questions for this modality, alongside activity, stability and specificity.^[S7] It is not an impurity problem or a formulation problem. It is a property of the sequence itself, and it is assessed by transcriptome-wide measurement rather than by checking a list of predicted matches, because the mechanism producing it does not require a full match.

The seed region problem, and why delivery cannot fix it

A short stretch at one end of the guide strand, the seed region, can pair with partially complementary sites in unintended transcripts and silence them the way a microRNA would. Recent work testing modified backbone chemistries in this region reports high tolerance of the modification while preserving on-target activity and mitigating these microRNA-like off-target effects.^[S9] The point for a development programme is that this is a sequence and chemistry problem, not a delivery problem. Improving the targeting ligand delivers the same off-target liability more efficiently. It is addressed by screening candidate sequences against the transcriptome early, when changing the sequence is still cheap, rather than after a lead has been selected.

Species differences in the sequence itself

This modality has a species problem that antibodies and small molecules do not. A sequence designed against the human transcript may not match the animal transcript at all, in which case the animal study measures the effect of an inactive oligonucleotide. The usual responses are to use a species where the sequences happen to match, or to make a separate surrogate sequence against the animal transcript and test that. The second option means the safety data describe a different molecule from the clinical candidate, and that limitation should be stated in the submission rather than left implicit.

The nonclinical work

Biodistribution

Plasma concentration is a poor guide here, because the material clears from circulation quickly while remaining active in tissue for far longer. What has to be measured is tissue concentration, and specifically concentration in the compartment where the machinery operates, since material sitting in an endosome is not material doing anything. For a particle formulation, the lipid components have their own distribution and clearance and are measured separately from the oligonucleotide.^[S2]

Confirming the mechanism

The direct readout is the amount of target transcript remaining, measured in the tissue of interest. Protein level is measured as well, since transcript reduction and protein reduction can differ in timing and extent depending on how long the existing protein persists. Where a circulating protein is the target, as with transthyretin, its serum concentration provides an accessible marker that can be followed in the same animal over time.^[S1][S5]

Model selection

Human hepatocyte systems, whether primary cells or derived from stem cells, are used because they carry the human transcript the candidate sequence was designed against and express the receptor a conjugate is directed at. What they cannot report is distribution, since there is no circulation. Animal studies supply that and are run with the species limitation described above stated explicitly. Neither substitutes for the other.

Safety assessment

Liver function, renal function and innate immune stimulation are the standard areas for this modality, with liver and kidney weighted by where the material accumulates. Off-target silencing belongs in the safety assessment rather than only in the discovery phase, because a transcript silenced unintentionally produces a toxicity finding with no obvious cause unless the transcriptome data exists to explain it. For a chronic product, the study duration has to be long enough for a slowly accumulating effect to appear.

What to establish with the laboratory

Four things are worth establishing before award. Whether the laboratory can quantify oligonucleotide in tissue, not only in plasma, and by what method. Whether it runs transcriptome-wide off-target screening and at what depth, given that this is a sequence problem rather than a formulation one.^[S9] How it would handle the species mismatch: whether it has experience qualifying a surrogate sequence, and what it considers adequate qualification. And, for a particle formulation, whether it measures the lipid components separately from the oligonucleotide.^[S2]

About this article

This is an independent editorial article for people who commission nonclinical work in the United States. Product descriptions and indications are taken from the prescribing information as published, and approvals are recorded as facts; this site does not discuss the efficacy of any product or compare products. Where two approved products differ, that difference is recorded without any claim that one is preferable. It does not recommend any laboratory. Last reviewed: September 3, 2026.

References

1. FDA Drugs@FDA / openFDA — ONPATTRO (patisiran)、NDA 210922. https://www.accessdata.fda.gov/scripts/cder/daf/ (accessed 2026-09-03)
2. FDA-approved labeling — ONPATTRO, Description. https://www.accessdata.fda.gov/scripts/cder/daf/ (accessed 2026-09-03)
3. FDA Drugs@FDA / openFDA — GIVLAARI (givosiran)、NDA 212194. https://www.accessdata.fda.gov/scripts/cder/daf/ (accessed 2026-09-03)
4. FDA-approved labeling — GIVLAARI, Description. https://www.accessdata.fda.gov/scripts/cder/daf/ (accessed 2026-09-03)
5. FDA Drugs@FDA / openFDA — AMVUTTRA (vutrisiran)、NDA 215515. https://www.accessdata.fda.gov/scripts/cder/daf/ (accessed 2026-09-03)
6. FDA-approved labeling — AMVUTTRA, Description. https://www.accessdata.fda.gov/scripts/cder/daf/ (accessed 2026-09-03)
7. Hu B, Zhong L, Weng Y . Therapeutic siRNA: state of the art. Signal Transduction and Targeted Therapy. https://pubmed.ncbi.nlm.nih.gov/ (accessed 2026-09-03)
8. Hepatocyte targeting via the asialoglycoprotein receptor. RSC Medicinal Chemistry, 2025. https://pubmed.ncbi.nlm.nih.gov/ (accessed 2026-09-03)
9. Acyclic serinol nucleic acid modification of siRNAs overcomes seed region mediated off-target effects. https://pubmed.ncbi.nlm.nih.gov/ (accessed 2026-09-03)