Part 5 — Therapeutics

19. The modality zoo — every way to lower a gene

"Modality" means what kind of medicine. Each attacks a different level of the central dogma from Module 1, and each has a distinct profile of speed, precision, durability, and risk.

flowchart TB
    DNA["<b>DNA</b><br/>the extra copy"] --> RNA["<b>mRNA</b><br/>extra working copies"] --> PROT["<b>Protein</b><br/>too much STAG2"]
    E1["<b>CRISPR excision</b><br/>cut out the duplication"] -.->|permanent| DNA
    E2["<b>CRISPRi / epigenetic silencing</b><br/>turn the copy down"] -.->|semi-permanent| DNA
    E3["<b>ASO</b> · <b>siRNA</b> · <b>Cas13 RNA editing</b><br/>destroy or reduce the transcripts"] -.->|reversible, tunable| RNA
    E4["<b>Small molecule</b><br/>indirect — alter regulation<br/>or downstream effects"] -.->|reversible, fast| PROT

The comparison that matters:

Reading this table for this condition. Recall from Module 4 that STAG2 is a tumor suppressor and the target is a partial reduction. That places unusual value on tunability and reversibility — the ability to adjust the dose, and to stop entirely if something goes wrong.

That reasoning favors the RNA-level modalities (ASO, siRNA, Cas13) over permanent DNA editing, despite DNA editing sounding like the more definitive "cure." A permanent edit that overshoots cannot be undone. This is a case where the more elegant-sounding solution is arguably the wrong one, and understanding why is a genuine mark of fluency.

The two-lane strategy this implies:

Both lanes run on the same engine — patient-derived cells. They are not competing strategies; they're a portfolio, and the durable lane must be started now precisely because it is slow.

Key terms:

• Modality — the class of therapeutic (pill, oligonucleotide, gene therapy…).
• ASO (antisense oligonucleotide) — short synthetic strand that binds and reduces a target mRNA.
• siRNA (small interfering RNA) — double-stranded RNA that triggers destruction of a target mRNA.
• RNAi (RNA interference) — the natural cellular mechanism siRNA exploits.
• CRISPRi — CRISPR machinery adapted to repress a gene rather than cut it.
• Knockdown vs knockout — partial reduction vs complete elimination. Knockdown is what a duplication needs.
• Off-target effects — unintended action at the wrong site.
• Conjugate — a chemical group attached to improve delivery (e.g. the divalent siRNA designs developed for CNS delivery).

References:

• n-Lorem Foundation — nonprofit creating free bespoke ASOs for nano-rare patients. Directly relevant; see Module 24.
• NCT06615206 — the HG204 RNA-editing trial

20. Getting a medicine into the brain

A therapy that cannot reach neurons cannot work, regardless of how good the dish data look. Delivery is often the binding constraint — and it's the step most often glossed over in optimistic conversations.

The blood-brain barrier (BBB) is a tight lining of blood vessels in the brain that excludes most large molecules. Small, fat-soluble molecules can sometimes cross. ASOs, siRNAs, and viruses essentially cannot — which is why they must be delivered directly.

Two consequences worth holding onto:

1. Route shapes a family's life, not just the science. "Effective but requires an intrathecal injection every four months for years" is a real, livable treatment — and a very different proposition from a daily pill. Both are wins; they're just different wins, and worth thinking about in advance.

2. BBB penetrance must filter hits early. In a repurposing screen, a compound that works beautifully in a dish but cannot reach the brain is a dead end. Filtering for CNS penetrance should happen early in triage (Module 18), not after everyone is emotionally invested.

Key terms:

• BBB (blood-brain barrier) — the selective barrier protecting the brain.
• Intrathecal — into the cerebrospinal fluid via the spine.
• ICV (intracerebroventricular) — into the brain's fluid-filled ventricles.
• CSF (cerebrospinal fluid) — the fluid bathing brain and spinal cord.
• Capsid — the protein shell of a virus; engineered capsids can improve brain targeting.
• CNS penetrance — whether a drug reaches the central nervous system at useful levels.
• Biodistribution — where in the body a drug actually ends up.

21. Safety — the constraints that shape everything

Constraint 1 — STAG2 is a tumor suppressor. This is the defining safety consideration, and it is not hypothetical. Loss of STAG2 is a recurrent event in several human cancers — bladder cancer, Ewing sarcoma, and some leukemias.

The implications are concrete:

• The therapeutic goal is a partial reduction (~30–50% is the figure commonly discussed), not maximal suppression.
• Overshoot is a genuine harm, not merely a lack of benefit — pushing toward MKMS on one axis and cancer risk on the other.
• Any long-term therapy needs long-term surveillance.
• This is a strong argument for brain-targeted delivery over whole-body exposure: the tumor-suppressor function matters most in dividing tissues, and neurons largely don't divide. Restricting the drug to the brain reduces exposure of the tissues where cancer risk actually lives. This is an elegant and important point — worth raising explicitly.

Constraint 2 — the dosage window is narrow. Both directions are harmful (Module 4). This demands dose-finding, monitoring, and ideally a reversible modality.

Constraint 3 — modality-specific risks:

Constraint 4 — a real human safety signal you should know about. In April 2026, Nature Medicine reported the first use of a non-allele-specific KCNT1 knockdown ASO in two infants with severe epileptic encephalopathy (Nakayama et al., PMID 41981306; senior author Timothy Yu, who developed milasen). Seizures were reduced — but both infants developed ventricular enlargement / hydrocephalus, and in one case goals of care were redirected.

This is the most important safety datum in this whole primer. It is not a reason to abandon the oligonucleotide route — it remains the most credible durable path. It is a reason that dose-finding, reversibility, imaging surveillance, and honest informed consent are non-negotiable, and that "ASOs are well tolerated" is too glib a summary of the current evidence. It should be raised explicitly in any advisory or consent discussion, not left as a footnote.

Constraint 5 — the child is not only a research subject. Every intervention carries opportunity cost: time, procedures, travel, missed school, and emotional load. A therapy that is modestly effective but highly burdensome may not be the right choice for your child. This is a family judgment, not a scientific one, and the family is the right party to make it.

Key terms:

• Tumor suppressor — a gene that restrains cancer development.
• Therapeutic window — the range between effective and harmful doses.
• Immunogenicity — the tendency to provoke an immune response.
• Pre-existing AAV immunity — prior natural exposure can render a person ineligible for AAV therapy.
• Surveillance — planned long-term monitoring for late effects.
• Risk–benefit assessment — the formal weighing that regulators and ethics boards require.

References:

• PubMed search: STAG2 tumor suppressor — the cancer literature, which defines the safety floor.