TL;DR
This paper introduces a novel 21-gene pair signature (21-GPS) that can accurately identify high-risk low-grade gliomas (LGGs) with molecular characteristics similar to aggressive glioblastomas (GBMs). This signature, based on relative gene expression orderings, offers a more robust classification than current methods, enabling earlier and more tailored treatment for patients with a poorer prognosis.
What Problem Does This Paper Address?
Gliomas, a type of brain tumor, are notoriously difficult to classify and predict outcomes for. While histopathological grading (e.g., Grade II, III, IV) exists, it suffers from subjective interpretation and interobserver variability, leading to misclassification and potentially suboptimal treatment. Molecular markers like IDH mutations and TERT promoter (TERT-p) mutations have improved classification, but their prognostic impact can be complex. Specifically, TERT-p mutations are associated with better outcomes in LGGs but worse outcomes in GBMs. This creates a critical challenge: identifying LGG patients who, despite their current grade, harbor GBM-like molecular features and are at high risk of progression, thus requiring more aggressive treatment. Current molecular classifications may miss these "GBM-like" LGGs, leading to under-treatment.
Methods
The researchers developed a molecular signature using a technique called within-sample relative expression orderings (REOs). Here's how it works:
- Differential Gene Expression: They first identified genes that were differentially expressed between TERT-p mutant LGGs and TERT-p mutant GBMs.
- Identifying Stable REOs: For each gene pair (Gene A and Gene B), they determined if the expression of Gene A was consistently higher than Gene B (A > B) or vice versa (A < B) across samples within a specific group (e.g., LGGs or GBMs). They looked for gene pairs with "significantly stable REOs," meaning this ordering occurred much more often than by chance (p < 0.01).
- Finding Reversely Stable Gene Pairs: The key innovation was identifying gene pairs that showed a stable REO in one direction in LGGs but the opposite stable REO in GBMs. These "reversely stable gene pairs" act as a molecular switch, indicating a shift in biological state.
- Developing the Signature: After rigorous filtering and selection to ensure consistency and avoid redundancy, they identified a signature composed of 21 such gene pairs (21-GPS).
- Classification Rule: For a new sample, the 21-GPS classifies it as "high-risk" if more than 10 of the 21 gene pairs exhibit the REO pattern characteristic of GBMs. Otherwise, it's classified as "low-risk." This majority voting approach makes the signature robust.
Datasets
The study utilized several publicly available datasets for training and validation:
- Training Datasets:
- TCGA (The Cancer Genome Atlas) - LGG and GBM cohorts.
- GEO (Gene Expression Omnibus) - Datasets GSE16011 and GSE61374 (specifically for low-grade samples).
- Validation Datasets:
- CGGA (Chinese Glioma Genome Atlas)
- GEO datasets GSE68848, GSE43388, and GSE16011 (for samples without TERT-p status information).
- E-MTAB-3892 (for initial analysis of TERT-p's ambivalent prognostic impact).
Key Insights and Ideas
- TERT-p's Bivalent Prognostic Role: The paper highlights the paradoxical effect of TERT-p mutations: beneficial in LGGs but detrimental in GBMs. This complexity necessitates a more nuanced approach than simple mutation status.
- GBM-like Signatures in LGGs: The core hypothesis that high-risk LGGs can share GBM-like molecular features is strongly supported. These tumors may be "silent" in terms of current grading but are on a trajectory towards malignancy.
- REOs as a Robust Classifier: Using relative gene expression orderings, rather than absolute expression levels or simple mutation presence/absence, provides a more stable and less sensitive signature to technical variations across different platforms (RNA-seq and microarray).
- Beyond IDH and TERT-p: The 21-GPS demonstrates superior prognostic value compared to IDH mutation status alone, especially in TERT-p wildtype samples, suggesting it captures a broader spectrum of aggressive tumor biology.
- Mutation Landscape Differences: The high-risk group identified by 21-GPS shows a distinct mutation profile, enriched for IDH wildtype and malignant factors like PTEN and EGFR, mirroring primary GBM characteristics, while low-risk groups resemble classic LGGs.
Key Results
- A 21-gene pair signature (21-GPS) was developed.
- The 21-GPS successfully classified LGGs into high-risk and low-risk groups with significantly different overall survival (OS).
- The high-risk group exhibited a similar genetic mutation profile and survival outcome to GBMs, indicating a propensity for progression.
- In TERT-p mutant LGGs, the 21-GPS identified a high-risk subgroup with significantly worse OS than TERT-p wildtype samples, and this OS was comparable to GBM samples.
- The 21-GPS showed higher prognostic value than IDH mutation status in stratifying LGGs, with higher hazard ratios and AUC values in ROC analyses.
- The 21-GPS proved to be an independent prognostic factor in multivariate Cox regression analyses across multiple datasets.
- The signature's performance was stable across different platforms (RNA-seq and microarray).
Limitations and Future Work
- Sample Size for Specific Subtypes: While the study used multiple datasets, the analysis of specific subtypes (e.g., oligodendrogliomas vs. astrocytomas within LGGs) might have limitations due to sample size in certain cohorts.
- Mechanism of TERT-p's Bivalent Effect: The paper acknowledges that the exact mechanism behind TERT-p's opposite prognostic impact in LGGs versus GBMs remains unexplained, suggesting this as an area for future research.
- Clinical Validation: While validated across multiple independent datasets, further prospective clinical validation is always a crucial next step for any diagnostic or prognostic biomarker.
- Integration with Other Markers: Future work could explore how the 21-GPS integrates with other emerging molecular markers or therapeutic targets to further refine treatment strategies.
- Understanding the 21 Genes: While the signature is functional, a deeper dive into the specific biological roles and interactions of the 21 genes within the signature could provide further mechanistic insights.
Summary updated