Computational tools for assessing gene therapy under branching process models of mutation

TL;DR

This paper develops a three-type branching process model to study mutation accumulation in stem cells post-gene therapy, focusing on the emergence of double-mutants with leukemogenic potential, and explores how mutation dynamics influence risks.

Contribution

It introduces a novel three-type branching process model for mutation accumulation in stem cells, providing quantitative tools to assess leukemia risk after gene therapy.

Findings

Increasing single-mutant numbers raises double-mutant probability.

Single-mutants can be proliferative without significantly increasing risk if initial insertion avoids mutants.

Model behavior varies with birth rates and initial population sizes.

Abstract

Multitype branching processes are ideal for studying the population dynamics of stem cell populations undergoing mutation accumulation over the years following transplant. In such stochastic models, several quantities are of clinical interest as insertional mutagenesis carries the potential threat of leukemogenesis following gene therapy with autologous stem cell transplantation. In this paper, we develop a three-type branching process model describing accumulations of mutations in a population of stem cells distinguished by their ability for long-term self-renewal. Our outcome of interest is the appearance of a double-mutant cell, which carries a high potential for leukemic transformation. In our model, a single-hit mutation carries a slight proliferative advantage over a wild-type stem cells. We compute marginalized transition probabilities that allow us to capture important quantitative aspects of our model, including the probability of observing a double-hit mutant and relevant moments of a single-hit mutation population over time. We thoroughly explore the model behavior numerically, varying birth rates across the initial sizes and populations of wild type stem cells and single-hit mutants, and compare the probability of observing a double-hit mutant under these conditions. We find that increasing the number of single-mutants over wild-type particles initially present has a large effect on the occurrence of a double-mutant, and that it is relatively safe for single-mutants to be quite proliferative, provided the lentiviral gene addition avoids creating single mutants in the original insertion process. Our approach is broadly applicable to an important set of questions in cancer modeling and other population processes involving multiple stages, compartments, or types.