Assessing the clinical benefit of a nomogram to predict specimen-confined disease at radical prostatectomy in patients with high-risk prostate cancer: An external validation

Abstract

ObjectiveThe objective of the current study was to test generalizability and clinical value of a recently published nomogram to predict specimen-confined disease (SCD, pT2-pT3a+R0+pN0) at radical prostatectomy (RP) in patients with clinical high-risk prostate cancer (HRPCa). The nomogram allows improved decision making with curative intent within this heterogeneous patient cohort, which is important, as RP in patients with clinical HRPCa remains a topic of controversy. MethodsWe externally validated the nomogram in 1,669 men with clinical HRPCa who underwent RP and extended pelvic lymph node dissection between 1992 and 2011. A Kaplan-Meier analysis to estimate 5- and 10-year biochemical recurrence–free survival was performed. To investigate the SCD model׳s performance, the previously reported regression coefficients of the SCD nomogram were applied. Within loess calibration plots, the extent of overestimation or underestimation was graphically explored. Finally, decision curve analysis (DCA) to assess the clinical value of the SCD nomogram was performed. ResultsOverall, 49% of men showed SCD after RP. The 5- and 10-year biochemical recurrence rates for men with SCD were 66% and 56%, respectively, vs. 32% and 20%, respectively, for men without SCD (log-rank test P<0.001). External validation demonstrated comparable accuracy in relation to accuracy derived from internal validation (68.1% vs. 72.0%). Calibration was suboptimal, showing a tendency to underestimate SCD probability. In DCA, the nomogram׳s usage was associated with a clinical net benefit relative to both treating all and none. ConclusionsWithin our cohort, the nomogram׳s use was associated with a clinical net benefit according to DCA. However, one-third of men were falsely classified as having SCD or non-SCD. Nevertheless, in the absence of superior tools, the SCD nomogram represents a useful clinical decision aid.