A novel algorithm for the prediction of prostate cancer in clinically suspected patients

Prostate cancer (PCa) represents the most common solid tumor affecting men and its early detection remains the best approach to improve survival rates. The assessment of serum levels of PSA is currently used for PCa screening but the low specificity of the test results in a high number of false positives. Other forms of PSA may be detected in the bloodstream including PSA associated with immunoglobulin M (PSA-IgM) which, alone or combined with PSA, has shown diagnostic accuracy for PCa.

The aim of the study is to improve the diagnostic accuracy of PSA-IgM by developing a multivariable model which includes serum biomarkers and routine diagnostic parameters to obtain a predictive index useful in the post-screening clinical practice.

One hundred sixty male patients with clinical suspect of PCa underwent a trans-rectal ultrasound guided first prostate biopsy with a standardized sampling scheme. To generate the model, we assessed the presence of PSA and PSA-IgM complexes in sera of patients and the prostate volume of each patient. A novel predictive probability for PCa (iXip) was obtained combining non-overlapping biomarkers normalized with diagnostic parameters.

The study population included 49 patients with PCa diagnosed at biopsy and 111 controls in which prostate biopsy showed the presence of benign prostatic hyperplasia, inflammation, atypical small acinar proliferation or high-grade prostatic intraepithelial neoplasia. The iXip values for patients with PCa (mean ± SD=0.467 ± 0.160) were significantly higher (p-value < 0.001) than control subjects (mean ± SD=0.314 ± 0.098) and the iXip AUC (0.787) was significantly greater (p-value < 0.001) than the AUCs of each biomarker.

iXip shows a significant increase in diagnostic performance compared to PSA and PSA-IgM and its post-screening use may facilitate decision-making in recommending for biopsy clinically suspected patients.