Computational modelling of cancer evolution by multi-type branching processes

TL;DR

This paper models cancer metastasis evolution using multi-type branching processes, focusing on mutation-driven resistance and providing numerical tools to analyze the timing of resistant cancer cell emergence.

Contribution

It introduces a numerical framework based on multi-type branching processes to study cancer evolution and resistance development, extending previous theoretical results.

Findings

Numerical algorithms for estimating mutation occurrence times

Analysis of lifespan distribution effects on cancer progression

Potential to improve understanding of treatment resistance dynamics

Abstract

Metastasis, the spread of cancer cells from a primary tumor to secondary location(s) in the human organism, is the ultimate cause of death for the majority of cancer patients. That is why, it is crucial to understand metastases evolution in order to successfully combat the disease. We consider a metastasized cancer cell population after medical treatment (e.g. chemotherapy). Arriving in a different environment the cancer cells may change their lifespan and reproduction, thus they may proliferate into different types. If the treatment is effective, in the context of branching processes it means, the reproduction of cancer cells is such that the mean offspring of each cell is less than one. However, it is possible mutations to occur during cell division cycle. These mutations can produce a new cancer cell type, which is resistant to the treatment. Cancer cells from this new type may lead to the rise of a non-extinction branching process. The above scenario leads us to the choice of a reducible multi-type age-dependent branching process as a relevant framework for studying the asymptotic behavior of such complex structures. Our previous theoretical results are related to the asymptotic behavior of the waiting time until the first occurrence of a mutant starting a non-extinction process and the modified hazard function as a measure of immediate recurrence of cancer disease. In the present paper these asymptotic results are used for developing numerical schemes and algorithms implemented in Python via the NumPy package for approximate calculation of the corresponding quantities. In conclusion, our conjecture is that this methodology can be advantageous in revealing the role of the lifespan distribution of the cancer cells in the context of cancer disease evolution and other complex cell population systems, in general.