Assessing the Risk of Prostate Cancer with Nutritional and Environmental Factors: A Cross-Sectional Study from...

Screening for prostate cancer (PCa) using prostate-specific antigen (PSA) is common. This study aimed to identify potential nutritional variables associated with PSA levels and develop a PCa risk prediction model. A total of 5,725 participants from the 2001-2010 National Health and Nutrition Examination Survey (NHANES) database were recruited and divided into discovery and validation sets. Three machine learning (ML) algorithms were performed in discovery set to select features associated with a high-risk PSA level. Age, red blood cell (RBC), blood cholesterol, blood lead, and prognostic nutritional index (PNI) were overlapping important contributors among the three ML models. Furthermore, the relationship between the selected variables and the high-risk PSA level was verified using weighted logistic regression models and restricted cubic spline (RCS) curves. After adjustment, age and blood lead were significantly positively associated with high-risk PSA, while PNI was significantly negatively associated with high-risk PSA. A nomogram including age, RBC, blood cholesterol, and PNI was constructed for predicting the PCa risk according to PSA levels, and the area under the receiver operating characteristic curve (AUROC) of discovery and validation sets was 0.755 and 0.756, respectively. The constructed nomogram may be used for targeted PSA testing and cancer screening in community healthcare.