Detecting the stable point of therapeutic effect of chronic myeloid leukemia based on dynamic network biomarkers

Junhua Xu, Min Wu, Shanshan Zhu, Jinzhi Lei, Jie GaoBMC Bioinformatics 2019 · DOI

TL;DR

This paper introduces a novel method using Dynamic Network Biomarkers (DNB) to pinpoint the "stable point" of therapeutic effect in chronic myeloid leukemia (CML) patients. By analyzing gene expression data, they developed a "Therapeutic Effect Index" (TEI) that can predict when a patient's condition has stabilized, offering a more precise way to assess treatment efficacy beyond traditional markers.

What Problem Does This Paper Address?

Chronic myeloid leukemia (CML) is a blood cancer where traditional biomarkers can only distinguish between a healthy state and a diseased state. However, they fall short in identifying the crucial "pre-stable state" during drug treatment. This means clinicians struggle to precisely determine when a patient's condition has stabilized after therapy, making it difficult to optimize treatment duration and predict outcomes. Current CML treatments, like tyrosine kinase inhibitors, are effective but can lead to resistance, treatment failure, and relapse. The paper aims to bridge this gap by developing a method to dynamically track and identify the point at which CML patients achieve a stable therapeutic response.

Methods

The researchers propose a strategy based on Dynamic Network Biomarkers (DNB). Here's a breakdown of their approach:

  • Data Preprocessing: They combined three public gene expression datasets and standardized them using Robust Multichip Averaging (RMA). Batch effects were corrected using the COMBAT method.
  • Differentially Expressed Genes (DEGs) Identification: A student's t-test was used to identify DEGs between control (healthy) and treatment groups, with a p-value cutoff of 0.05 and a fold change of 1.5.
  • DNB Identification: DEGs were clustered using hierarchical clustering. Based on three criteria proposed by Chen, a group of 250 genes was selected as the core DNB.
  • Therapeutic Effect Index (TEI) Construction: To quantify the therapeutic effect and identify the pre-stable state, they developed the TEI. This index is based on three dynamic measures of the DNB:
  • Coefficient of Variation (CV): Measures the variability of molecules within the DNB group.
  • Pearson Correlation Coefficient (PCC): Measures the correlation between genes within the DNB group.
  • Outward PCC (OPCC): Measures the correlation between DNB genes and genes outside the DNB group. The TEI is calculated as CVt * PCCt / OPCCt.

  • Functional Analysis: Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses were performed on the identified DNB genes to understand their biological roles in CML.

Datasets

The study utilized three publicly available gene expression datasets from the National Center for Biotechnology Information's Gene Expression Omnibus (GEO) database:

  • GSE33075
  • GSE12211
  • GSE24493

These datasets were combined to create a larger dataset for analysis, resulting in a total of 39 chips. The CML diagnosed samples were used as the control group.

Key Insights and Ideas

  • Dynamic Biomarkers for Dynamic States: The core idea is that disease progression and treatment response are dynamic processes, and static biomarkers are insufficient. DNBs, by capturing these dynamic changes, can reveal subtle shifts in biological systems.
  • The "Pre-Stable State" is Detectable: The paper demonstrates that there's a discernible period before a patient's condition fully stabilizes during treatment. Identifying this "pre-stable state" is crucial for understanding treatment dynamics.
  • TEI as a Predictive Tool: The developed TEI effectively tracks the therapeutic response over time, showing a clear trend towards a stable, healthy state. This index can serve as a quantitative measure of treatment efficacy.
  • Key Genes Identified: The DNB analysis highlighted specific genes (e.g., BCR, TGFBR2, ABL1, CRK) that are strongly implicated in CML pathogenesis and progression, suggesting potential new therapeutic targets.

Key Results

  • A DNB consisting of 250 genes was identified.
  • Out of these, 43 genes were found to be closely related to CML pathogenesis.
  • The Therapeutic Effect Index (TEI) showed a clear trend of decreasing over time, approaching a healthy baseline after 1 month of imatinib treatment. This indicates that 1 month is the optimal time for curative effect.
  • The study found that at the pre-stable state, the OPCC value was not as prominent, while CV and PCC values showed significant changes, aligning with theoretical expectations.
  • Functional enrichment analysis revealed that DNB genes are involved in critical pathways like hematopoietic cell lineage, cytokine-cytokine receptor interaction, apoptosis, MAPK signaling pathway, and PI3K-Akt signaling pathway, all relevant to CML.

Limitations and Future Work

  • Limited Sampling Points: The authors acknowledge that the study had a limited number of sampling points for data collection, which made it challenging to fully predict the pre-stable state.
  • Future Directions: The researchers plan to focus on refining their algorithm and further investigating the prediction of the pre-stable state in future work. They also suggest that the identified biomarkers could help in prompt treatment, minimizing drug resistance, treatment failure, and relapse, thereby reducing CML mortality. The identified genes could also serve as new therapeutic targets.

Summary updated

BibTeX
@article{xu2019detecting,
  author = {Junhua Xu and Min Wu and Shanshan Zhu and Jinzhi Lei and Jie Gao},
  title = {{Detecting the stable point of therapeutic effect of chronic myeloid leukemia based on dynamic network biomarkers}},
  journal = {BMC Bioinformatics},
  year = {2019},
  doi = {10.1186/s12859-019-2738-0},
}
Figure