Identification of Pathogenesis and Prognoses of Relapse in Myeloma by Bioinformatic Method

Haoshu Zhong, Yang Liu, Jialin Duan, Xiaohong Chen, Hao Xiong, Kunyu Liao, Chunlan HuangResearch 2021 · DOI

Abstract

Multiple Myeloma (MM) is the second most common hematological malignancy, characterized by a persistent cycle of remission and relapse. The molecular mechanisms driving this recurrence remain poorly understood. This study utilizes bioinformatic methods, including Weighted Correlation Network Analysis (WGCNA) and CIBERSORT, to identify key gene modules and immune cell interactions associated with MM progression. We identified a two-gene prognostic model based on TPX2 and PRAM1 that effectively predicts event-free survival (EFS). Furthermore, our analysis highlights the critical role of the bone marrow microenvironment, specifically the interaction between M2 macrophages and cytokine signaling, in fostering a permissive landscape for relapse.

Introduction

Despite significant therapeutic advances, Multiple Myeloma remains largely incurable due to the high frequency of drug-resistant relapses. Understanding the biological "black box" of why and when these relapses occur is essential for improving patient outcomes. This research aims to move beyond traditional clinical markers by identifying the underlying molecular pathways and gene regulatory networks that drive MM progression.

Methodology

The study integrated gene expression and clinical data from the Gene Expression Omnibus (GEO, GSE24080) and The Cancer Genome Atlas (TCGA).

  • Network Analysis: WGCNA was employed to identify modules of co-expressed genes correlated with clinical traits.
  • Immune Profiling: The CIBERSORT algorithm estimated the proportions of 22 immune cell types to assess the tumor microenvironment.
  • Prognostic Modeling: Univariate and multivariate Cox regression analyses were used to develop a risk-scoring model based on key prognostic genes.

Functional Insights: The "Yellow" Module and Cell Cycle Regulation

WGCNA identified three primary modules—yellow, brown, and sky-blue—most closely associated with clinical relapse. The "yellow" module was significantly enriched for functions related to the cell cycle and DNA replication.

Functional classification of dysregulated genes associated with Multiple Myeloma relapse.

As shown in the table above, central players such as AURKB and CENPA are involved in critical processes like mitosis and cytokinesis. The dysregulation of these genes suggests that MM relapse is driven by a fundamental "re-awakening" of the cell's proliferative machinery.

The Immune Microenvironment and Cytokine Signaling

The brown and sky-blue modules were linked to cytokines and their receptors. Analysis revealed a significant positive correlation between M2 macrophage infiltration and cytokines like CCL18, CCL2, and CXCL12. Furthermore, cytokines such as CCL8 and CCL23 were found to correlate positively with plasma cell fractions. This suggests that the tumor microenvironment is actively remodeled into a "pro-tumor" soil that supports survival and drug resistance, facilitating recurrence.

A Two-Gene Prognostic Model

The study narrowed down hundreds of candidates to a robust two-gene signature: TPX2 (a spindle assembly factor) and PRAM1.

The established risk score is calculated as follows: $$Risk Score = (1.48739 \times \text{expression of } TPX2) + (0.62980 \times \text{expression of } PRAM1)$$

In the training set, this model achieved an Area Under the Curve (AUC) for predicting 2-year, 3-year, and 4-year event-free survival (EFS) of 0.717, 0.643, and 0.624, respectively. Multivariate analysis confirmed that this risk score is an independent predictor of survival, providing prognostic value beyond traditional markers like LDH and B2M.

Conclusion

This bioinformatic study provides a novel framework for predicting MM relapse and identifies key molecular drivers of disease progression. While the statistical associations are strong, future "wet lab" validation is required to confirm the functional roles of TPX2 and PRAM1. Additionally, targeting the identified cytokine pathways (e.g., CCL18) may offer new therapeutic avenues to disrupt the supportive microenvironment that enables MM recurrence.

Summary updated

Abstract

Abstract Background: Multiple myeloma (MM), the second most hematological malignancy, the molecular mechanism and pathogenesis of the relapse of MM is poorly understood. This study aimed to identify novel prognostic model for MM and explore potential mechanism of relapse. Methods: Gene expression data,clinical data(GSE24080) and HTseq-Counts files were downloaded from Gene Expression Omnibus (GEO) and TCGA database. Co-expression modules of genes were built by Weighted Correlation Network Analysis (WGCNA).KEGG and GO enrichment analysis were performed in each module. TATFs (tumor-associated transcription factors) were retrieved from the Cistrome. Twenty-two immune cell compositions was calculated by CIBERSORT algorithm.Univariate and multivariate Cox congression were performed and a predictive model by prognostic genes was constructed,the predictive power of the model was evaluated by Kaplan–Meier curve and time-dependent receiver operating characteristic (ROC) curves. Results: A total of 940 DEGs were identified,and in WGCNA analysis, yellow, brown and sky-blue modules were most associated with clinic traits.The yellow module related with the cell cycle and the brown and sky-blue modules correlated with cytokine and its receptors, where the M2 macrophage fraction is positively correlated with CCL18, CCL2, CCL8, CXCL12 and CCl23 were positively correlated with plasma cells by Cibersort analysis.Prognostic genes were identified and two genes (TPX2,PRAM1) were finally identified to construct a risk model for predicting EFS.
BibTeX
@inproceedings{zhong2021identification,
  author = {Haoshu Zhong and Yang Liu and Jialin Duan and Xiaohong Chen and Hao Xiong and Kunyu Liao and Chunlan Huang},
  title = {{Identification of Pathogenesis and Prognoses of Relapse in Myeloma by Bioinformatic Method}},
  booktitle = {Research},
  year = {2021},
  doi = {10.21203/rs.3.rs-1100874/v1},
}
Figure