Identification and construction of a novel NET-related gene signature for predicting prognosis in multiple myeloma

Abstract

Neutrophil extracellular traps are essential in the development and advancement of multiple myeloma (MM). However, research investigating the prognostic value with NET-related genes (NRGs) in MM has been limited. Patient transcriptomic and clinical information was sourced from the gene expression omnibus database. Cox regression analysis with a univariate approach was employed to explore the link between NRGs and overall survival (OS). Kaplan-Meier methods were applied to assess variations in survival rates. A nomogram integrating clinical data and predictive risk metrics was crafted using multivariate logistic and Cox proportional risk model regression analyses. Additionally, we investigated the disparities in biological pathways, drug sensitivity, and immune cell involvement, and validated differential levels of two key genes through qPCR. We identified 148 differentially expressed NRGs through published articles, of which 14 were associated with prognosis in MM. Least absolute shrinkage and selection operator Cox regression model established a nine-gene NRG signature-comprising ANXA1, ANXA2, ENO1, HIF1A, HSPE1, LYZ, MCOLN3, THBD, and FN1-that demonstrated strong predictive power for patient survival. The Cox regression model with multiple variables demonstrated that the risk score independently predicted OS, showing that those with a high score had worse survival rates. Furthermore, a nomogram incorporating patient age, LDH levels, the International Staging System, and NRGs was developed, demonstrating strong prognostic prediction capabilities. Drug sensitivity correlation analysis also offered valuable guidance for future immuno-oncological therapies and drug selection in MM patients. The NRGs signature was a reliable biomarker for MM, effectively identifying high-risk patients and forecasting clinical outcomes.

Keywords: Multiple myeloma; Neutrophil extracellular traps; Prognostic signature; Tumor prognostic biomarkers.

Fig. 1

Determination of prognostic NRGs and construction of prognostic models. A Forest plots of univariable Cox model. B LASSO coefficient profiles of the expression of ten candidate genes. C Selection of the optimal gene in the Lasso model (λ)

Fig. 2

Prognostic analysis of the eight-gene signature in the training cohort. A The distribution and median value of the risk scores in the training cohort. B The distributions of OS status, OS time, and risk score in the training cohort. C PCA plots of the GEO database training cohort population. D t-SNE analysis of the training cohort population. E Kaplan–Meier curves of OS for patients in the high-risk and low-risk groups of the training cohort. F UC of time-dependent ROC curves verified the prognostic performance of the risk score in the training cohort

Fig. 3

Validation of the eight-gene signature in the validation cohort. A Distribution and median values of risk scores in the validation cohort. B Validate the distribution of OS status, OS time, and risk score in the queue. C PCA plots of the validation cohort population. D t-SNE analysis of the validation cohort population. E Kaplan–Meier curves of OS for patients in the high-risk and low-risk groups of the validation cohort. F AUC of time-dependent ROC curves in the validation cohort. G Kaplan–Meier curves of OS for stage II MM patients in the training cohort. H Kaplan–Meier curves of OS for stage II MM patients in the validation cohort

Fig. 4

Results of the univariate and multivariate Cox regression analyses regarding OS and construction of a nomogram. A Results of univariate Cox regression analysis of the GSE2658 training cohort. B Results of multivariate Cox regression analysis of the GSE2658 training cohort. C Univariate Cox regression analysis of the GSE136337 validation cohort. D Multivariate Cox regression analysis of the GSE136337 validation cohort. E Construment of the nomogram. F Calibration plot used to predict the 1-, 3-, and 5-year survival. G–I ROCs for 1-, 3-, and 5-year eigenfactors in the training cohort. J–L ROC for 1-, 3-, and 5-year eigenfactors in the validation cohort

Fig. 5

Comparison of the ssGSEA scores between different risk groups. A Scoring of sixteen immune cell types in the training cohort. B Twelve immune-related functions of the GEO cohort. (aDC, Activated dendritic cell; iDC, Immature dendritic cell; pDC, Plasmacytoid dendritic cell; Tfh, T follicular helper cell; Th2, T helper 2; TIL, Tumor infiltrating lymphocyte; Treg, Regulatory T Cell; HLA, Human leukocyte antigen; APC, Antigen presenting cell; CCR, Cytokine–cytokine receptor; Adjusted p values were showed as: *, p < 0.05, **, p < 0.01, ***, p < 0.001.)

Fig. 6

Functional analysis of the gene expression profile between the NRGs low- and high-risk groups. A GO enrichments analysis of upregulated genes in the high-risk group. B GO enrichments analysis of upregulated genes in a low-risk group. C KEGG enrichment analysis of upregulated genes in the high-risk group. D KEGG enrichment analysis of upregulated genes in a low-risk group

Fig. 7

Scatterplot of the relationship between prognostic gene expression and drug sensitivity

Fig. 8

Relationship between risk scores and treatment sensitivity

Fig. 9

Relative mRNA levels of the two NRGs were assessed by qPCR. A HSPE1. B MCOLN3. *p < 0.05, **p < 0.01, ***p < 0.001