Developing and validating a model for the PSA, Age and BMI relationships

INTRODUCTION:  Previous studies indicate an inverse relationship between prostate-specific antigen (PSA) and body mass index (BMI), and a positive relationship between PSA and age. We develop and validate a model relating PSA to age and BMI, and derive approximate PSA threshold values for prostate biopsy adjusted for BMI and age.

METHODS:  Cross sectional analysis nested within the ProtecT (Prostate testing for cancer and Treatment) trial. Of 18,238 ProtecT men aged 50-69 years, 9,457 men met inclusion criteria (no history of prostate cancer or diabetes, PSA < 10 ng/ml, BMI between 15 and 50 kg/m²). We analysed data from 7,527 men randomly assigned to a development dataset, and validated the model on 1,930 men in a validation dataset. Multivariable linear regression models were used to investigate the relationship between PSA, age and BMI.

RESULTS:  Mean PSA was 1.36 ng/ml (SD: 1.16), mean age was 61.6 years (SD: 5.0) and mean BMI was 27.20 kg/m² (SD: 3.76). In the multivariable model there was an 8.38% increase in PSA per 5 year increase in age (95% CI: 7.25% to 8.51%) and a 3.23% decrease in PSA per 5 kg/m² increase in BMI (1.74% to 4.72%). We used this model to derive an expression to adjust PSA for age and BMI, and to generate adjusted threshold values for prostate biopsy. 

CONCLUSIONS:  Age and BMI were associated with small changes in PSA in men without diabetes. The use of an equation model to adjust PSA was most relevant for men with borderline PSA values.