Optimizing the clinical utility of PCA3 to diagnose prostate cancer in initial prostate biopsy

BMC Cancer

Research Article

Ana Calatrava¹  Gerardo Sanz²  Francisco Martínez³  Angel Borque⁴  Miguel Ramírez-Backhaus⁵  Jose Domínguez-Escrig⁵  Inmaculada Iborra⁵  Argimiro Collado⁵  Alvaro Gómez-Ferrer⁵  Jose Rubio-Briones⁵  Juan Casanova⁵  Luis M. Esteban⁶  Jose A. Lopez-Guerrero⁷  Antonio Fernandez-Serra⁷  Irene Casanova-Salas⁷  Luis Rubio⁷ 

[1] Department of Pathology, Instituto Valenciano de Oncología, Valencia, Spain;Department of Statistical Methods, Universidad de Zaragoza, Zaragoza, Spain;Department of Statistics, University of Valencia, Valencia, Spain;Department of Urology, Hospital Universitario Miguel Servet, Zaragoza, Spain;Department of Urology, Instituto Valenciano de Oncología, C/ Prof. Beltrán Báguena 8, 46009, Valencia, Spain;Escuela Universitaria Politécnica de La Almunia, Universidad de Zaragoza, Zaragoza, Spain;Laboratory of Molecular Biology, Instituto Valenciano de Oncología, Valencia, Spain;

关键词: Probability Density Function;    Receiver Operating Characteristic;    Prostate Volume;    Threshold Probability;    Discrimination Ability;   

DOI  :  10.1186/s12885-015-1623-0

 received in 2015-04-21, accepted in 2015-08-21,  发布年份 2015

来源: Springer

【 摘 要 】

BackgroundPCA3 has been included in a nomogram outperforming previous clinical models for the prediction of any prostate cancer (PCa) and high grade PCa (HGPCa) at the initial prostate biopsy (IBx). Our objective is to validate such IBx-specific PCA3-based nomogram. We also aim to optimize the use of this nomogram in clinical practice through the definition of risk groups.MethodsIndependent external validation. Clinical and biopsy data from a contemporary cohort of 401 men with the same inclusion criteria to those used to build up the reference’s nomogram in IBx. The predictive value of the nomogram was assessed by means of calibration curves and discrimination ability through the area under the curve (AUC). Clinical utility of the nomogram was analyzed by choosing thresholds points that minimize the overlapping between probability density functions (PDF) in PCa and no PCa and HGPCa and no HGPCa groups, and net benefit was assessed by decision curves.ResultsWe detect 28 % of PCa and 11 % of HGPCa in IBx, contrasting to the 46 and 20 % at the reference series. Due to this, there is an overestimation of the nomogram probabilities shown in the calibration curve for PCa. The AUC values are 0.736 for PCa (C.I.95 %:0.68–0.79) and 0.786 for HGPCa (C.I.95 %:0.71–0.87) showing an adequate discrimination ability. PDF show differences in the distributions of nomogram probabilities in PCa and not PCa patient groups. A minimization of the overlapping between these curves confirms the threshold probability of harboring PCa >30 % proposed by Hansen is useful to indicate a IBx, but a cut-off > 40 % could be better in series of opportunistic screening like ours. Similar results appear in HGPCa analysis. The decision curve also shows a net benefit of 6.31 % for the threshold probability of 40 %.ConclusionsPCA3 is an useful tool to select patients for IBx. Patients with a calculated probability of having PCa over 40 % should be counseled to undergo an IBx if opportunistic screening is required.

【 授权许可】

CC BY   
© Rubio-Briones et al. 2015