UTILIZING MACHINE LEARNING TO UNCOVER SIGNATURE GENES AND REGULATORY MECHANISMS IN PITUITARY TUMORS: INSIGHTS FROM ATHLETIC PATIENTS

Abstract

Background: Pituitary tumors, though rare within the central nervous system tumor spectrum, have seen an uptick in incidence rates due to advancements in diagnostic screenings. Frequently misdiagnosed due to symptom overlap with more common conditions, these tumors can lead to severe complications, diminishing both the life expectancy and quality of life of affected individuals. This risk is exacerbated in athletic patients, where pituitary tumors can significantly impact physical performance and overall health, and where the adverse effects of conventional treatments like chemotherapy and radiotherapy may be particularly debilitating. Identifying new biomarkers is thus crucial for early diagnosis and the development of more targeted therapies. Methods: This study utilized datasets from the GEO database to integrate and analyze pituitary tumor-related data. Initially, differential expression gene (DEG) screening along with GO, KEGG, and GSEA enrichment analyses were conducted. Subsequently, LASSO and SVM-RFE algorithms were employed to pinpoint signature genes associated with pituitary tumors in a training set, followed by ROC performance verification and gene expression difference assessments in a test set. Immune infiltration discrepancies were analyzed, focusing on the correlation between signature genes and immune cell presence, with a particular interest in variations found in athletic patients. Results: Six signature genes—CNTNAP2, LHX3, RAB11FIP3, SOX9, TBX19, and TGFBR—were identified, each showing distinct expression patterns linked to the abundance of tumor-infiltrating immune cells. These findings suggest a notable interplay between the identified signature genes and the immune environment within pituitary tumors, highlighting potential pathways for immunologically targeted interventions. Conclusion: The discovery of six signature genes in pituitary tumors opens new avenues for the early diagnosis and treatment of this condition, particularly in athletic patients who may suffer disproportionate impacts from conventional treatments. These genes lay the groundwork for developing immune-targeted therapeutic strategies, offering hope for more effective and less invasive options for managing pituitary tumors.