Presentation Authors: Nicola Fossati*, Milan, Italy, R. Jeffrey Karnes, Stephen A. Boorjian, Luca Boeri, Rochester, MN, Alberto Bossi, Villejuif, France, Nadia Di Muzio, Cesare Cozzarini, Barbara Noris Chiorda, Elio Mazzone, Francesco Barletta, Giorgio Gandaglia, Milan, Italy, Detlef Bartkowiak, Ulm, Germany, Dirk Böhmer, Berlin, Germany, Shahrokh Shariat, Gregor Goldner, Vienna, Austria, Antonino Battaglia, Steven Joniau, Gaetan Devos, Karin Haustermans, Gert De Meerleer, Leuven, Belgium, Valérie Fonteyne, Piet Ost, Ghent, Belgium, Hein Van Poppel, Leuven, Belgium, Francesco Montorsi, Milan, Italy, Thomas Wiegel, Ulm, Germany, Alberto Briganti, Milan, Italy
Introduction: Previous studies defined biochemical response (BR) to salvage radiotherapy (SRT) as a first PSA after SRT < 0.2 ng/ml. However, the entity of early PSA response may be a more sensitive end-point of clinical progression as shown in the setting of systemic treatments. We hypothesized that the optimal definition of BR to SRT vary according to PSA at SRT.
Methods: We included 450 patients treated with local radiation to the prostatic and seminal vesicle bed for PSA rising after radical prostatectomy. The definition of BR was based on the first post-SRT PSA. No patient received concomitant HT. To combine the impact of pre- and post-SRT PSA levels, we defined &[Prime]PSA decrease ratio&[Prime] (PSADR): ((PSA at SRT - First PSA after SRT) / PSA at SRT)*100. The outcome was clinical recurrence (CR), defined by positive imaging at follow-up. Cox regression analysis was used to develop a base model that included pT stage, pathological Gleason score, and PSA at SRT. The optimal PSADR cut-off was defined using the multivariable function Lowess. Two BR definitions were compared: (1) post-SRT PSA < 0.2 ng/ml; (2) optimal PSADR cut-off. Decision curve analysis provided an estimate of the net benefit obtained using the two definitions (Model 1 vs Model 2). The accuracy of the models was assessed using the AUC.
Results: Median (IQR) PSA pre-SRT was 0.40 (0.22, 0.69) ng/ml. Median (IQR) PSA post-SRT was 0.10 (0.02, 0.23) ng/ml. Median (IQR) PSADR was 75% (44, 94). At a median follow-up of 79 months, 67 (15%) patients developed CR. Using the Lowess function, we identified PSADR >0% as the optimal cut-off for BR (Figure 1a), which equals to any post-SRT PSA drop of at least 0.1 ng/ml compared to the pre-treatment PSA. We observed BR in 329 (73%) patients using definition #1 (post-SRT PSA < 0.2 ng/ml), and in 396 (88%) patients using definition #2 (PSADR >0%). At decision curve analysis, Model 2 showed higher clinical net benefit compared to Model 1(Figure 1b). The AUC of Model 2 was higher as compared to Model 1 (AUC 75% vs. 71%, p=0.001).
Conclusions: Any PSA drop â‰¥0.1 ng/ml in the first PSA after SRT represented the most accurate definition of BR to SRT. This new definition out-performed the current definition based on post-SRT PSA only and should be validated in future studies.