Steven Cen

Background/Objectives: Needle biopsy-based molecular assays that are sensitive, cost-effective, and reproducible could improve prostate cancer (PCa) diagnosis and risk stratification. We developed a compact, biologically informed gene expression signature panel (ONECUT2, EZH2, BRCA2) for cancer detection and aggressiveness stratification. Methods: We designed and validated a multiplex real-time RT-qPCR assay for ONECUT2, EZH2, and BRCA2. Biomarker selection was guided by The Cancer Genome Atlas Prostate Adenocarcinoma (TCGA-PRAD) and Gene Expression Omnibus (GEO) cohort analyses. The assay was applied to 979 ex vivo needle biopsy cores from 255 radical prostatectomy specimens; machine-learning models (Random Forest, AdaBoost, Elastic Net) generated an RNA signature. Diagnostic and prognostic performance was assessed by the area under the receiver operating characteristic curve (AUC), model comparisons by DeLong’s test, and biochemical recurrence-free survival (BCR) by Kaplan–Meier and Cox proportional hazards analyses. Reproducibility based on sample pairs was assessed by intraclass correlation coefficients (ICC). Results: The three-gene expression signature (ONECUT2, EZH2, and BRCA2) robustly discriminated malignant from non-malignant tissue (AUC = 0.91) and identified clinically significant cancer (Gleason Grade Group [GG] ≥ 2 vs. GG 0–1, AUC = 0.89) within biopsy cores. However, it had limited performance for advanced patient-level features (e.g., pathological GG3–5, AUC = 0.63). In contrast, ONECUT2 gene-body methylation significantly stratified pathological aggressiveness (pathological GG3–5, AUC = 0.79; aggressive disease, AUC = 0.75) and demonstrated superior assay stability (methylation ICC 0.58–0.70 vs. RNA ICC 0.38–0.61). Furthermore, ONECUT2 DNA methylation significantly stratified BCR-free survival (hazard ratio = 3.90, p = 0.023), whereas the RNA signature showed a modest trend (hazard ratio = 1.88, p = 0.053). Notably, integrating DNA methylation with the three-gene expression signature (combined model) significantly improved predictive accuracy for adverse features (e.g., extraprostatic extension, ΔAUC = 0.20, p = 0.02) compared to the three-gene expression signature alone. Conclusions: While a multiplex RT-qPCR panel (ONECUT2, EZH2, and BRCA2) excels at core-level cancer detection, it lacks the resolution for aggressiveness classification. In contrast, ONECUT2 DNA methylation provides superior analytical reproducibility, robust stratification of cancer aggressiveness, and long-term recurrence risk estimation. These findings suggest that integrating transcriptional and epigenetic markers into a multimodal framework could provide complementary information to refine risk stratification in PCa.

Background Metabolic dysfunction-associated steatotic liver disease (MASLD) is affected by metabolic, psychological, and behavioral factors, with potential differences in pathophysiological pathways across racial groups. Objective This study aims to investigate the relationship between eating behaviors, emotional functioning, and anthropometric differences in Asian and maintain at various stages of liver disease. Methods A retrospective analysis was conducted on 98 MASLD patients (70 White and 28 Asian) from a Southern California liver clinic. Participants completed validated questionnaires assessing depressive and anxiety symptoms, eating behaviors, and quality of life. Body composition was measured using bioelectrical impedance analysis, and liver health was assessed through transient elastography. Multivariate regression and mediation analyses examined associations and pathways among psychological, behavioral, and clinical variables, with race as a moderator. Results White participants had higher body mass index (BMI) and body fat percentage, but liver steatosis and stiffness were similar between groups. In Asians, hepatic steatosis correlated most strongly with body fat mass, whereas in Whites, emotional eating was more closely linked to BMI and body fat percentage compared to Asians. Whites reported higher anxiety and lower quality of life. Emotional eating was associated with hepatic steatosis in both groups, but only in Whites did BMI and percent body fat mediate this relationship. Among Asians, emotional eating was linked to liver fat independent of body composition. Conclusion Distinct racial pathways exist in MASLD progression, with Asians potentially developing fatty liver without overt body fat changes. These findings highlight the need for tailored diagnostic and clinical strategies that consider metabolic, behavioral, and psychological differences across racial groups.

Clear cell renal cell carcinoma (ccRCC) is the most common type of renal cancer. Extensive efforts have been made to utilize radiomics from computed tomography (CT) imaging to predict tumor immune microenvironment (TIME) measurements. This study proposes a Green Learning (GL) framework for approximating tissue-based biomarkers from CT scans, focusing on the PD-L1 expression and CD68 tumor-associated macrophages (TAMs) in ccRCC. Our approach includes radiomic feature extraction, redundancy removal, and supervised feature selection through a discriminant feature test (DFT), a relevant feature test (RFT), and least-squares normal transform (LNT) for robust feature generation. For the PD-L1 expression in 52 ccRCC patients, treated as a regression problem, our GL model achieved a 5-fold cross-validated mean squared error (MSE) of 0.0041 and a Mean Absolute Error (MAE) of 0.0346. For the TAM population (CD68+/PanCK+), analyzed in 78 ccRCC patients as a binary classification task (at a 0.4 threshold), the model reached a 10-fold cross-validated Area Under the Receiver Operating Characteristic (AUROC) of 0.85 (95% CI [0.76, 0.93]) using 10 LNT-derived features, improving upon the previous benchmark of 0.81. This study demonstrates the potential of GL in radiomic analyses, offering a scalable, efficient, and interpretable framework for the non-invasive approximation of key biomarkers.

Radiomics involves extracting quantitative features from medical images, resulting in high-dimensional data. Unsupervised clustering has been used to discover patterns in radiomic features, potentially yielding hidden biological insights. However, its effectiveness depends on the selection of dimensionality reduction techniques, clustering methods, and hyperparameter optimization, an area with limited exploration in the literature. We present a novel bootstrapping-based hyperparameter search approach to optimize clustering efficacy, treating dimensionality reduction and clustering as a connected process chain. The hyperparameter search was guided by the Adjusted Rand Index (ARI) and Davies–Bouldin Index (DBI) within a bootstrapping framework of 100 iterations. The cluster assignments were generated through 10-fold cross-validation, and a grid search strategy was used to explore hyperparameter combinations. We evaluated ten unsupervised learning pipelines using both simulation studies and real-world radiomics data derived from multiphase CT images of renal cell carcinoma. In simulations, we found that Non-negative Matrix Factorization (NMF) and Spectral Clustering outperformed the traditional Principal Component Analysis (PCA)-based approach. The best-performing pipeline (NMF followed by K-means clustering) successfully identified all three simulated clusters, achieving a Cramér’s V of 0.9. The simulation also established a reference framework for understanding the concordance patterns among different pipelines under varying strengths of clustering effects. High concordance reflects strong clustering. In the real-world data application, we observed a moderate clustering effect, which aligned with the weak associations to clinical outcomes, as indicated by the highest AUROC of 0.63.