AAV vector genomes (vg) delivered to a hepatocyte mostly stay episomal โ circular DNA sitting in the nucleus, not integrated into a chromosome. Episomes don't have their own replication machinery, so when a hepatocyte divides, its episomes are not duplicated first. Each existing copy is inherited by only one of the two daughter cells:
Cell with k copies divides โ
daughter A gets Binomial(k, 0.5) copies
daughter B gets the remaining (k โ that) copies
E[copies per daughter] = k / 2 (dilution by half, in expectation, per division)
After g divisions: mean copies โ kโ ยท (1/2)แต
This simulator models a small lobule of hepatocytes (each an instanced sphere). At the chosen division rate, a random subset of cells "divides" every tick: their episomes are split binomially between the cell and a randomly chosen neighbour standing in for the sibling that replaces a naturally turned-over cell, keeping the lobule's cell count constant while still diluting the vector pool exactly as real hepatocyte turnover would.
- Division rate โ models hepatocyte turnover. A quiescent adult liver turns over only a few percent of cells per year; a growing pediatric liver, or one recovering from injury, turns over far faster.
- Trigger Liver Regeneration Event โ a one-off burst of ~35% of the lobule dividing at once, mimicking a partial hepatectomy or acute liver injury.
- Relative transgene expression โ assumed roughly proportional to mean vector-genome copies per cell, matching the clinical observation that AAV-delivered clotting-factor levels in hemophilia gene therapy decline over years in a pattern that tracks vector-copy dilution, not immune clearance.
Real-world relevance: this is a documented limitation of first-generation AAV liver gene therapy (e.g. hemophilia A/B trials) โ durability is markedly better in adults with low hepatocyte turnover than in children whose livers are still growing, and a regenerative insult (surgery, viral hepatitis, heavy alcohol use) can accelerate the same dilution in an adult.