Machine Learning-Driven PCDI Classifier for Invasive PitNETs

Guanyu Wang ^1,2, Song Yan ^1,2, Luyang Zhang ¹

⁰Department of Neurosurgery, The Second Affiliated Hospital of Harbin Medical University, Harbin, 150086, China.

Current Gene Therapy +

|

July 7, 2025

View abstract on PubMed

Summary

A novel Programmed Cell Death Index (PCDI) accurately distinguishes aggressive pituitary tumors. This biomarker captures immune-metabolic crosstalk, offering new avenues for personalized therapy in invasive PitNETs.

Area Of Science

Oncology
Molecular Biology
Immunology

Background

Aggressive Pituitary Neuroendocrine Tumors (PitNETs) present treatment challenges due to invasiveness and therapy resistance.
Existing prognostic markers fail to capture molecular heterogeneity, highlighting the need for novel biomarkers.
Dysregulated Programmed Cell Death (PCD) pathways are implicated in cancer, but their role in invasive PitNETs is unclear.

Purpose Of The Study

To identify novel molecular biomarkers for aggressive PitNETs.
To investigate the prognostic relevance of PCD pathways in invasive PitNETs.
To develop a predictive index for risk stratification and personalized therapy.

Main Methods

Differential gene expression analysis of GEO datasets (GSE51618, GSE169498, GSE260487) comparing noninvasive and invasive PitNETs.
Integration of a 1,548-gene PCD-related gene panel.
Machine learning (LASSO, SVM-RFE) to construct a PCD-associated Index (PCDI); validation via ROC analysis, immune infiltration assessment, and RT-qPCR.

Main Results

The 11-gene PCDI accurately differentiated invasive from noninvasive PitNETs.
High-PCDI tumors showed enriched metabolic pathways and immune activation.
Consensus clustering identified two subtypes; C2 (high-PCDI) exhibited increased immune scores and pathway activity, with key gene expression validated experimentally.

Conclusions

The PCDI surpasses traditional models by integrating PCD-immune-metabolic crosstalk for improved prognostic accuracy in invasive PitNETs.
High-PCDI tumors display immune evasion despite checkpoint molecule expression, suggesting combined MAPK inhibitor and immunotherapy potential.
The PCDI offers a molecular framework for risk stratification and personalized treatment of invasive PitNETs.

Keywords:

Pituitary neuroendocrine tumors gene expression omnibus immune infiltration machine learning pan-cancer. prognostic models programmed cell death-associated index