Self-organizing maps to aid prognostic and diagnostic biomarker identification in exploratory metabolomics of benign prostatic hyperplasia.

INTRODUCTION

Population aging is increasing rapidly, representing a global trend that has launched many discussions regarding older adult's quality of life. Benign prostatic hyperplasia (BPH) is a commonly found condition among older men, affecting the lower urinary tract and quality of life. BPH diagnosis typically involves clinical evaluations which cannot accurately identify patients who will develop a clinically significant disease in the future. Metabolomics emerges as a promising approach to understand the metabolic pathways associated with BPH, offering potential aid in early diagnosis.

OBJECTIVES

To investigate the metabolic profile of BPH patients compared with a control group for diagnosing BPH at different stages.

METHOD

A total of 62 individuals were selected and divided into two groups: 32 BPH (prostate volume > 40 mL) and 30 healthy individuals (prostate volume ≤ 40 mL and normal voiding patterns). Plasma samples were analyzed by LC-HRMS. Data were processed using unsupervised and supervised techniques. Self-organizing maps were further used to aid in the identification of potential biomarkers.

RESULTS

Self-organizing maps revealed three distinct groups which expanded upon the existing groups indicated by the medical team. Discriminant analysis models showed good predictive ability and accuracy, providing a classification perspective of patients and expanding diagnostic possibilities for different health states. Integration of chemical data with clinical variables revealed meaningful correlations. Our results also suggest the presence of discriminatory metabolites related to inflammation and oxidation processes, clarifying the metabolic basis of BPH.

CONCLUSION

Self-organizing maps proved their effectiveness in classifying groups of samples from BPH patients, revealing potential biomarkers that went undetected by conventional data analysis methods. This approach underscores metabolomics as a relevant tool in identifying hidden patterns and supporting the development of more robust diagnostic and prognostic models.