Safety and Peri-Operative Outcomes During Learning Curve of Robot-Assisted Laparoscopic Prostatectomy: A Multi-Institutional Study of Fellowship-Trained Robotic Surgeons Versus Experienced Open Radical Prostatectomy Surgeons Incorporating Robot-Assisted Laparoscopic Prostatectomy

Introduction

T here is no consensus regarding the proper training or credentialing required before a surgeon performs robot-assisted laparoscopic prostatectomy (RALP). The responsibility of determining when a particular surgeon is qualified to perform RALP often rests with local hospital administrations. The number of urologists incorporating RALP into practice is steadily increasing and the training background for these surgeons is becoming more varied and diverse as the popularity of RALP continues to grow. An emerging issue centers on the number of cases required to overcome the learning curve associated with RALP. Unfortunately, most studies pertaining to the learning curve of RALP are single surgeon or institution reviews that often fail to address the relevant issue of background training or experience.

A subset of fellowship-trained robotic surgeons (FEL) has emerged into the field, and it is logical to assume that given their focused training these surgeons will progress quickly to proficient robotic prostatectomists. More specifically, their progression should be abbreviated when compared with experienced open surgeons (OE) incorporating RALP. Rosser et al ¹ previously studied the impact of fellowship training on the outcomes of radical retropubic prostatectomy (RRP). The authors reported that FEL progressed to comparable results of more experienced surgeons within their first year of practice. In the present study, we elected to evaluate the results of fellowship training on safety and peri-operative outcomes of RALP. Explicitly, we compare RALP outcome metrics between two FEL and three OE using the first 30 consecutive cases for each surgeon. To better understand the effect of the natural learning curve for OE, we also examined the next consecutive set of 30 cases for each OE. Our rationale for selecting 30 cases as a reference point is based on findings from previous studies that reported a substantial improvement in surgical margin status and surgical outcomes after the initial 30 RALP. ²

Materials and Methods

After Institutional Review Board approval at each institution, we prospectively collected data on the initial 30 RALP cases for two FEL robotic surgeons. The two FEL were at different institutions but completed the same laparoscopic and robotic fellowship in consecutive years at the same institution. The fellowship is a 1-year program and is credentialed by the Endourological Society and recognized by the American Urologic Association. During the program, each FEL completed portions of over 45 RALP under the supervision of one experienced mentor. This information was compared with the first 30 RALP each of three experienced RRP surgeons (OE) (defined as at least 1000 RRP in their careers) who had incorporated RALP into their practice. The OE all had experience with hand-assisted laparoscopic kidney removal.

The OE performed RALP after a training course provided by Intuitive Surgical Corporation (Sunnyvale, CA) and were proctored appropriately for their first cases. The training course occurred for 2 days and included both cold (models) and warm simulation (pig dissection). In addition to our primary analysis comparing the first 30 cases from the three OE with the first 30 cases form the two FEL, we also compared the second consecutive 30 cases from the OE with the first 30 from the FEL to account for improvements that may occur in the OE after some level of experience is gained with the technology.

RALP was performed in a similar manner to the Vattikuti Institute procedure. ³ Operative technique was similar between the surgeons and institutions. Both centers utilized the original da Vinci surgical platform. Laparoscopic port sites were comparable between the two sites. Anastomotic techniques were similar between the two sites (double-armed Monocryl suture in a running van Velthoven fashion). ⁴ Suture and needle type were similar between the two sites. Foley catheter drainage was maintained between 7 and 14 days. Surgical assistance was provided by urology residents at different levels of training. Assistants were variable throughout the study period.

Outcomes were evaluated from final pathological specimens, perioperative data, and the 3-month postoperative return visit. Outcome data included length of hospital stay, hemoglobin change, duration of catheter use, duration of surgical drain, hospital readmission, and mortality. Hospital stay and surgical drains were considered prolonged if >2 days in duration. Foley catheter removal time and pre-removal cystogram requirement was at the discretion of the operating surgeon. Catheter time was considered prolonged if it was in place >14 days. Failure to reach a prostate-specific antigen (PSA) nadir <0.15 ng/mL was recorded. Primary Gleason score of 4 or greater or a total Gleason score of 8 or greater was used to define high-grade cancer. A positive margin was defined as tumor at the inked margin upon microscopic assessment. Patients requiring a re-operation, defined as exploratory laparotomy, cystoscopy, or urethral dilation, were recorded as well. Data were compared for each group utilizing nonparametric Wilcoxon tests for continuous variables and Chi-square test for categorical variables. Three separate groups were compared: group A, the first 30 cases from the FEL; group B, the first 30 cases from the OE; group C, the second 30 cases from the OE. Median values were reported.

Results

Complete data were available for 236 patients (4 patients had incomplete data sets and were not included in the analysis). There were no rectal injuries in either group. There was no death reported during the study follow-up period. No lymphoceles, incisional hernias, or wound dehiscence were reported in either group. Table 1 details baseline characteristics of the study population. The initial 30 cases for each surgeon in the FEL group and the OE group were statistically similar with regard to body mass index, PSA, and Gleason score. During the first 30 cases the FEL group was less likely to operate on clinical stage T2 disease than the OE group (p = 0.003). The number of patients with palpable disease operated on by the OE group decreased significantly in the second 30 cases.

Table 2 compares peri-operative data and outcomes between the initial 30 FEL cases, the initial 30 OE cases, and the next 30 OE cases. FEL were more likely to complete a pelvic lymph node dissection (PLND) than the OE in their first 30 cases (p = 0.043) and their second 30 cases (p = 0.051). The PLND included obturator lymph node sampling. Extended pelvic lymph node sampling was not completed. For the first 30 cases, the FEL group was less likely to have prolonged urethral catheter drainage (5% vs. 19%; p = 0.009) and require re-operation (0% vs. 8%; p = 0.042). One patient from the OE cohort was re-explored for hemorrhage 6 hours after surgery and subsequently found to have an occult bleeding disorder. The remaining re-operations consisted of cystoscopy and urethral dilation. During the first 30 cases, the FEL had a total positive margin rate of 15% (9/60) compared with a 34% (31/90) margin positive rate in the initial 30 OE procedures (p = 0.008). One patient in the FEL group did not reach a PSA nadir <0.15 ng/mL within 3 months of surgery compared with 9 out of 90 (10%) in the OE group (p = 0.056). When we compared the second 30 cases from the OE with the cases from the FEL, the OE group moved into statistically comparable rates to FEL in all measures except re-operation and the performance of PLND.

Table 3 displays a direct comparison of the OE with regard to the first 30 cases and the second 30 cases. Positive margin rates improved from 34% to 19%, failure of to nadir at undetectable levels at 3 months declined from 10% to 3%, and patients requiring prolonged catheter drainage improved from 19% to 6%. There were three cases requiring open conversion in the OE group during their first 30 surgeries. These were after respiratory distress secondary to pneumoperitoneum and failure to progress. The FEL group and the OE group in their second 30 cases required no open conversions. All hospital readmissions were for postoperative ileus that was treated conservatively.

Discussion

Incorporation of new technology into surgical practice leads to concern with regard to morbidity, mortality, and overall patient safety. As individual hospitals attempt to credential surgeons for RALP, it is important to have persistent and repeatable data on its initial use among surgeons. Early investigations reported that only 12 to 20 cases were required to gain RALP proficiency. ^3,5 In contrast, Herrell and Smith ⁶ felt that comparable RALP outcomes to RRP were not achieved until over 150 RALP had been completed. The same publication commented that surgeon comfort and confidence did not fully manifest until 250 cases. Regardless of training level or experience, surgeons from this multicenter study demonstrated an excellent safety profile with no death, no rectal injuries, and only a single re-exploration over their first 236 cases. Open conversion was a rare event in the study and was not present in the fellowship-trained group. There were three open conversions for the OE group that occurred only during the first 30 cases. Six patients (3%) from the overall study population required blood transfusion (3%), consistent with published RALP data and further confirming a known benefit of RALP compared with RRP. ^7,8 Previous studies have documented typical hospital stay after RALP to be between 1 and 2 postoperative days. ^{5,7,9 –11} Only 8% of the study population regardless of surgeon experience was hospitalized beyond that duration.

Surgeons learning RALP note the vesicourethral anastomosis to be one of the most difficult portions of the operation to master. ^12,13 Vesicourethral anastomosis takes longer for residents and fellows to gain proficiency compared with any other portion of RALP. ¹³ OE incorporating RALP have noted difficulty completing the vesicourethral anastomosis and have proposed numerous techniques to prevent leakage, including Lapra-Ty clips. ¹² Consistent with this, the OE group had a much higher rate of prolonged catheterization compared with the FEL group in the first 30 cases. This improved dramatically in the second 30 cases for the OE group and approached the rate of the FEL group with experience. The increased Foley catheter time secondary to urinary leakage present in the OE group may account for the much higher rate of re-operation in the form of cystoscopy and urethral dilation present in the OE group. In fact, no patient in the FEL group required re-operation. It is assumed that the FEL group had gained anastomotic proficiency within their fellowship year.

Consistent data on 5-year PSA recurrence-free survival after RALP are still lacking. Surgical margin status continues to be used as a surrogate for cancer control in initial RALP series. RRP series have reported positive margins ranging between 16% and 46%. ¹⁴ Reported RALP margin positivity has ranged between 3% and 35.5%. ^{2,15 –18} White et al ¹⁹ noted a 22% positive margin rate for the first 50 RALP cases of an experienced RRP surgeon. This rate was lower than the surgeon's previous 50 RRP. In our current study, both groups of surgeons were within the reported positive margin range. The FEL demonstrated a clear advantage with regard to margin status during their first cases. While the OE were initially more than twice as likely to have positive margins, this disparity diminished as did failure of PSA to nadir at undetectable levels over the course of the second 30 cases. In a study of 500 RALP cases Patel et al ¹⁵ found that 5% of patients had a detectable PSA over a 10-month follow-up period. The fellowship-trained cohort in our study had one patient who failed to nadir at undetectable PSA levels (1%). The open surgeons were able to move into this range within the course of this investigation decreasing from 10% to 3% in their second 30 cases. Long-term follow-up is required to insure a durable absence of PSA recurrence.

A consistent finding within this study was that the FEL were more likely than OE to perform PLND. This was in the setting of a higher amount of palpable disease in the OE group and no difference in risk based on Gleason score. If all patients with PSA over 10 and/or biopsy Gleason 4 ± 3 disease were considered high risk for lymph node dissection purposes, 100% (10/10) of these patients received PLND in the FEL group. The PLND rate in the initial 30 high risk patients in the OE group was only 28% (5/18). This increased to 57% (8/14) in the second 30 cases for the OE group. This may be attributable to the extent of fellowship training or the comfort level of the FEL performing PLND. Further case numbers for the OE would be required to see if this is only related to experience with the procedure. Reported operating room time is variable in RALP reports and was not evaluated in our study. It is possible that the OE group required more time to perform their initial RALP and this precluded their ability to perform subsequent PLND.

The authors acknowledge that this study has several limitations. There were some dissimilarities noted in the baseline demographics with respect to palpable disease and previous surgical histories. Although not likely, we cannot discount those differences potentially affecting the operative outcomes among the groups. Follow-up was limited to 3 months in this study as our main goal was to asses peri-operative safety and morbidity of RALP. Also, our data support the concept that surgeons who undergo fellowship training in robotic surgical techniques have better short-term outcomes for their initial RALP cases compared with OE who initiate RALP into their practice. While the premise is logical, we are mindful that a primary assumption of our study design is that there are no other differences between our two groups of surgeons, other than their fellowship training, that could also be accounting for the observed difference in surgical outcomes. As an example, one might argue that while we have attributed the differences we observed in positive margin rates to be solely because of fellowship training status, there could be some other unidentified difference between the two groups of surgeons that resulted in the disparate rates. While we cannot completely rule out this possibility, we would point out that when we compared the later surgeries performed by the OE to the FEL, the difference in outcomes for most of the endpoints we measured was noticeably attenuated. This supports the original notion that the difference in outcomes we observed is primarily attributable to experience with the robotic technique, which is the primary benefit of fellowship training. This concept is further supported by the well-documented effect of experience on outcome improvement in the RRP literature. ^20,21 It should also be noted that the OE continued to perform RRP during their incorporation of robotic technology and the cases were not consecutive. Cases were chosen by surgeon comfort level. The FEL group performed all the cases consecutively. This also may serve as a marker of the FEL group's initial comfort level with RALP compared with the OE group. This study serves as baseline for assessing the impact of fellowship training on the morbidity of RALP.

The authors acknowledge the importance of functional outcomes such as continence and potency with regard to prostate cancer treatment. Since the focus of this article is on safety and peri-operative outcomes, that data are not reported herein. The open surgeons incorporating RALP into their practice had each performed over 1000 RALP in their careers. Their familiarity with prostatectomy in general may have contributed to their early safety profile. Two groups absent from this evaluation are RRP surgeons without the above level of experience and trainees just completing residency. Further, a cohort of non-fellowship-trained residents might serve as a baseline to compare the advantages of both substantial open experience as well as fellowship training.

Conclusions

Surgeons from a variety of training backgrounds can safely introduce RALP into their practices. OE can incorporate RALP safely and achieve peri-operative results similar to FEL within 60 cases. Improved peri-operative outcomes early in the RALP experience can be expected after fellowship training.