Advertisement

Predicting progression in active surveillance; external validation of the Canary PASS risk calculator with the Spanish Urological Association Registry on AS

Login to Access Video or Poster Abstract: MP43-12
Sources of Funding: NONE

Introduction

The Spanish Urological Association Registry on Active Surveillance (AS) [AEUPIEM/2014/0001, piem.aeu.es?_x000D_ NCT 02865330, clinicaltrials.gov] has comparable design and similar aims to the recently published Canary Prostate Active Surveillance Study (PASS, NCT00756665, clinicaltrials.gov). _x000D_ An easy to use on line predictive tool of progression in FU biopsy (Bx) has_x000D_ been proposed by the PASS group, the PASSRisk Calculator (PASSRC). We perform an external validation of the PASSRC in our series, specially focused on clinical utility of PASSRC by selecting cutoff points of probability for clinical decision counselling._x000D_

Methods

After matching for validation purposes, we select 498 patients with a minimum of one follow-up Bx? no other exclusion criteria were considered nor bias has been detected. PASSRC external validation is done by means of calibration curve and area under de ROC curve (AUC), identifying cutoffs of clinical utility by probability density functions (PDF) and clinical utility curves (CUC).

Results

We find significant differences in age, PSA and clinical stage between our validation cohort and the PASSRC generation cohort (p<.0001), with a progression rate of 10-22% on the successive follow-up Bx. No cancer was found in 44% of the first followup Bx. The calibration curve shows underestimation of observed progression. The AUC is 0.65 (C.I.95%: 0.60-0.71). PDF and CUC do not suggest a specific cutoff point for clinical use, because of the overlap of the distributions of probabilities between progressing and non-progressing patients_x000D_ _x000D_

Conclusions

In the first external validation of the PASSRC we have obtained a moderate discrimination ability. Unfortunately, we cannot recommend cutoff_x000D_ points of clinical use from our study. Specific risk calculators from different cohort features, different prognosis models or the inclusion of new biomarkers and/or morphofunctional parameters from mpMRI should be stressed as potential predictors for progression within AS strategies_x000D_

Funding

NONE

Authors
Angel Borque-Fernando
Jose Rubio-Briones
Luis M. Esteban-Escaño
Argimiro Collado-Serra
Ana Soto
Pedro A. López González
Jordi Huguet Pérez
Jose I. Sanz Velez
Jesús Gil Fabra
Enrigue Gómez Gómez
Cristina Quicios Dorado
Lluís Fumadó
Sara Martinez-Breijo
Juan Soto Villalba
In behalf of rest of AI of AEU/PIEM/2014/0001
back to top