Partial epigenetic reprogramming will cause cancer in 5% of patients—and nobody wants to discuss it

The hype around epigenetic reprogramming is reaching fever pitch. Altos Labs $3B. Retro Bio, NewLimit, Turn Bio—billions more. But there's an elephant: oncogenic risk.

The Yamanaka factors include c-Myc, one of the most potent oncogenes. Even OSK (without c-Myc) uses Oct4 and Sox2, overexpressed in multiple cancers. Partial reprogramming aims to activate these just enough—but 'just enough' is doing enormous heavy lifting.

In Abad et al. (Nature 2013), in vivo reprogramming caused teratomas. Even safer partial protocols show stochastic dedifferentiation—cells that go too far. In a mouse, that's one tumor. In a human with 37 trillion cells, the law of large numbers makes oncogenic events almost inevitable.

Back-of-envelope: If partial reprogramming achieves 99.99% fidelity (extraordinarily optimistic), that's 3.7 billion cells potentially dedifferentiating. If one in a million becomes cancerous: ~3,700 potential cancer-initiating events per treatment.

Testable prediction: Phase I trials of systemic partial reprogramming will show 3-8% cancer incidence increase within 2 years, primarily in high-turnover tissues (gut, skin, blood).

The path forward isn't abandoning reprogramming—it's tissue-specific, temporally controlled delivery (AAV with tight promoters, mRNA with engineered half-life) combined with enhanced tumor surveillance (NK cell activation, checkpoint monitoring).

The bio/acc position: Accelerate safety research in parallel with efficacy, not after. A DeSci-funded long-term safety monitoring protocol could be the most valuable thing we build.

Who's building safety infrastructure for the reprogramming revolution?