Clustering Multi-Omics Data in Gleason Scores $3+4$ and $4+3$ Reveals 3 Patterns

The Gleason Score (GS) is widely used to classify prostate cancer (Pca), with Gleason 7 (GS7) tumors typically being classified into Gleason <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$3+4$</tex> and Gleason <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$4+3$</tex>. However, this classification may overlook other molecular subtypes, limiting risk stratification and personalized treatment. Using multi-omics data analysis and unsupervised machine learning, this study investigates GS7 omics patterns. A total of 341 prostate cancer samples were clustered using five clustering techniques - <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$k-\text{Means}$</tex>, hierarchical clustering, Gaussian Mixture Model (GMM), DBSCAN, and Mean Shift Clustering, and cluster validity was determined using PCA visualization, dendrograms, and silhouettes. To further validate the biological relevance of the clusters, supervised classification models, including Random Forest and Support Vector Machine (SVM), were employed, achieving high classification accuracy (95.50% with Random Forest, 97.30% with SVM). GS7 tumors consistently showed three distinct subtypes, supporting the hypothesis of an unidentified molecular group. This finding challenges conventional classification paradigms and has significant implications for risk assessment and treatment strategies in prostate cancer. Future studies integrating larger patient cohorts and clinical validation are essential to establish the translational significance of this novel subtype.