Learning curve of minimally invasive radical prostatectomy: Comprehensive evaluation and cumulative summation analysis of oncological outcomes

Background and objective The primary objective was to evaluate the learning curve of minimally invasive radical prostatectomy (MIRP) in our institution and analyze the salient learning curve transition points regarding oncological outcomes. Methods Clinical, pathologic, and oncological outcome data were collected from our prospectively collected MIRP database to estimate positive surgical margin (PSM) and biochemical recurrence (BCR) trends during a 15-year period from 1998 to 2013. All the radical prostatectomies (laparoscopic prostatectomy [LRP]/robot-assisted laparoscopic radical prostatectomy [RARP]) were performed by 9 surgeons. PSM was defined as presence of cancer cells at inked margins. BCR was defined as serum prostate-specific antigen >0.2 ng/ml and rising or start of secondary therapy. Surgical learning curve was assessed with the application of Kaplan-Meier curves, Cox regression model, cumulative summation, and logistic model to define the “transition point” of surgical improvement. Results We identified 5,547 patients with localized prostate cancer treated with MIRP (3,846 LRP and 1,701 RARP). Patient characteristics of LRP and RARP were similar. The overall risk of PSM in LRP was 25%, 20%, and 17% for the first 50, 50 to 350, and>350 cases, respectively. For the same population, the 5-year BCR rate decreased from 30% to 16.7%. RARP started 3 years after the LRP program (after approximately 250 LRP). The PSM rate for RARP decreased from 21.8% to 20.4% and the corresponding 5-year BCR rate decreased from 17.6% to 7.9%. The cumulative summation analysis showed significantly lower PSM and BCR at 2 years occurred at the transition point of 350 cases for LRP and 100 cases for RARP. In multivariable analysis, predictors of BCR were prostate-specific antigen, Gleason score, extraprostatic disease, seminal vesicle invasion, and number of operations (P<0.05). Patients harboring PSM showed higher BCR risk (23% vs. 8%, P< 0.05). Conclusions Learning curve trends in our large, single-center experience show correlation between surgical experience and oncological outcomes in MIRP. Significant reduction in PSM and BCR risk at 2 years is noted after the initial 350 cases and 100 cases of LRP and RARP, respectively.