Background

The Evolutionary Dynamics of Cancer and the Challenges of Co-mutation Analysis

Cancer is not caused by a single gene defect but is a complex disease involving multiple genes. Various mutations within a patient interact with each other, driving the evolution of cells into malignant tumors. Therefore, researchers have focused on co-mutation analysis, which examines the occurrence of two or more mutations simultaneously, and mutual exclusivity analysis, which investigates whether the presence of a specific mutation excludes the presence of other mutations. These interactions are key to understanding the molecular mechanisms by which cancer cells acquire survival advantages and drug resistance.

Existing statistical analysis methods have limitations in controlling for various factors that influence tumor mutation burden (TMB), such as differences in TMB among patients, as well as mutational factors like smoking and ultraviolet exposure. Normal tissues also accumulate random mutations with age, making it difficult to distinguish mutation combinations that trigger cancer cell formation through simple frequency comparisons. As a result, a significant amount of false-positive information is not filtered out, leading to considerable confusion in the process of discovering new drug targets.

Key Findings

Development of the Computational Framework SelectSim and Analysis of 70,000 Genomic Maps

The research team at the University of Lausanne (UNIL), led by Professor Giovanni Ciriello, developed SelectSim, a computational framework designed to eliminate these confounding factors and identify true evolutionary dependencies. SelectSim is designed to precisely simulate the probability that gene pairs will mutate together by chance, reflecting each sample's unique mutation profile and TMB. Based on this, the team constructed a simulation-based null model and compared it with the observed frequencies to determine statistical significance.

The results of a large-scale comparative analysis of more than 70,000 cancer patient samples and normal tissue genomic data, covering 119 cancer subtypes, are intriguing. Specific co-mutation combinations that drive the malignant transformation of cancer cells are rarely found in healthy normal tissues and are only found in cancer tissues. This shows that even if random mutations accumulate in normal cells due to aging, they do not develop into malignant tumors unless specific co-occurring mutations occur. Furthermore, the research team observed that the composition of co-mutation combinations continuously changes from the initial stage of tumor formation to the primary tumor stage and to the metastatic stage, where the tumor spreads to other organs. In the evolutionary process of the tumor, the genetic alliances that cancer cells choose for survival and metastasis are constantly changing.

Significance and Prospects

Potential for Precise Metastasis Prediction and the Challenge of Data Dependency

This study is praised for precisely identifying the genetic rules that determine cancer evolution by removing noise from simple random mutations in cancer genomic big data. With the completion of a map of co-mutations according to the stage of evolution, it is possible to predict and block the cancer metastasis pathways that vary from patient to patient.

However, the predictive power of the machine learning model presented by SelectSim is inevitably dependent on the quality and quantity of the input genomic data. In the case of rare cancers where data acquisition is difficult, there are limitations in constructing a precise statistical model. In addition, functional follow-up studies are needed to demonstrate the molecular biological interactions that the identified mutation combinations induce in cells.

Why It Matters

This technology can be applied to the development of precision diagnostic chips or companion diagnostic technologies that can predict cancer metastasis early in the clinical setting. For example, when a specific single mutation is found in the tissue examination of a stage 1 colorectal cancer patient, it is difficult to accurately predict the prognosis. However, if the metastasis-inducing co-mutation combination discovered by SelectSim is also detected, a scenario can be designed to administer high-intensity chemotherapy proactively to increase the cure rate. In the drug development stage, it can also be used as a powerful filter to analyze gene combinations that induce drug resistance and select optimal combination therapy candidates. As a result, it is expected to increase the efficacy of anticancer drugs and effectively reduce the probability of treatment failure, accelerating the realization of personalized precision medicine.