Immune cell profiles and predictive modeling in osteoporotic vertebral fractures using XGBoost machine learning...

Area of Science:

• Osteoimmunology
• Computational Biology
• Bone Metabolism

Background:

• Osteoporosis leads to vertebral fractures (VF), impacting quality of life, especially in postmenopausal women.
• Fractures are often undiagnosed and influenced by bone density and mechanical loads.
• Immune cell interactions with bone cells are vital for bone repair and understanding bone pathologies.

Purpose of the Study:

• Investigate immune cell profiles in osteoporotic patients with and without vertebral fractures.
• Utilize xCell signatures and XGBoost predictive modeling.
• Identify immune biomarkers associated with vertebral fracture risk.

Main Methods:

• Employed xCell for immune cell signature analysis.
• Applied XGBoost algorithm for predictive modeling of vertebral fractures.
• Conducted gene expression analysis for osteoclast and osteoblast markers.

Main Results:

• Vertebral fracture patients showed increased CD4+ naïve T cells and central memory T cells.
• XGBoost model identified Th1 cells, CD4 memory T cells, and hematopoietic stem cells as key VF predictors.
• VF patients had reduced Th1 cells, enriched Th17 cells, upregulated osteoclast genes, and downregulated osteoblast genes.

Conclusions:

• Immune cell dysregulation, particularly involving T cell subsets, plays a critical role in osteoporotic vertebral fracture pathogenesis.
• XGBoost modeling effectively identifies immune biomarkers for vertebral fracture risk.
• Findings suggest novel therapeutic targets within immune pathways for fracture prevention in osteoporosis.