Introduction

Live donor nephrectomy is an elective procedure, where the patient undergoes surgery for the sole benefit of another individual. It is a procedure with minimal margin for error and the preferred outcome would be no change to the donor’s preoperative condition. Even in the best case scenario, the patient may still experience pain, the cosmetic effects of the procedure, and loss of time from work. As such, the safety of the patient is of utmost importance, and every endeavor should be made to minimize risks and improve the donation experience.

Ratner et al. [1] first introduced laparoscopic donor nephrectomy (LDN) in 1995. LDN demonstrated several improvements over open donor nephrectomy, including decreased postoperative pain, decreased length of hospital stay, faster recuperation, and reduced perioperative blood loss [2]. Due to these advantages, LDN is currently the standard of care, and several modifications have been made to improve the technique. These modifications include hand-assisted LDN, laparoendoscopic single-site surgery, transvaginal LDN, and robot-assisted laparoscopic donor nephrectomy (RDN) [3,4,5,6]. The introduction of LDN as a minimally invasive procedure with decreased recovery time and better cosmetic results also brought about an initial increase in the number of available live kidney donors [7]. However, this trend has now reversed as there has been a recent decrease in living kidney donations from 6647 in 2004 to 5633 in 2016 [8]. With the advent of robotic surgery, RDN may be the next evolution of the surgical technique for donor nephrectomy as well as a potential solution to the decreasing donor pool.

The DaVinci Surgical System (Intuitive Surgical, Mountain View, CA, USA) was first introduced in 2000. Since then, multiple procedures have incorporated the assistance of the system [9,10,11,12,13]. Robotic surgery is now considered to be the preferred approach or even the standard of care in multiple procedures including partial nephrectomy and radical prostatectomy [14,15,16,17]. The surgical robot has been viewed as “enabling technology” as it enables surgeons to perform complex minimally invasive procedures, which may be facilitated by the greater degrees of freedom allowed [18]. Despite surgeons generally having less experience with robotic surgery, it has been shown that robotic cases have achieved equivalent outcomes compared to the conventional cases for perioperative variables, including blood loss and transfusion rate [19]. By decreasing the learning curve for difficult surgical tasks, surgical robots may also expand the number of available surgeons for complex procedures as well as allow newer surgeons to quickly master these procedures.

Robot-assisted laparoscopic donor nephrectomy (RDN) was first reported by Horgan [6]. They showed that RDN retained the cosmetic advantages of the standard LDN, but also allowed for seven degrees of freedom and 3D vision in comparison with the four degrees of freedom and 2D vision of LDN [6]. With these benefits, RDN can serve as a potential alternative to LDN. However, the efficacy and safety of the procedure, including long-term postoperative outcomes, have yet to be thoroughly examined. In this study, we describe our institution’s initial experience with RDN by retrospectively comparing intraoperative and postoperative outcomes of donors undergoing RDN and LDN, performed by a single surgeon.

Materials and methods

The study commenced after approval from the Icahn School of Medicine at Mount Sinai Institutional Review Board (IRB). Informed consent was waived by the IRB. Data, including operative and postoperative parameters, were collected from 73 standard laparoscopic donor nephrectomies and 22 robot-assisted laparoscopic donor nephrectomies performed by a single surgeon from 2011 to 2016. Live donor demographic data included age, sex, race, body mass index (BMI), ASA status, and kidney laterality. Operative parameters included estimated blood loss, operative time, warm ischemia time (WIT), and extraction time. Warm ischemia time was defined as the time from clamping of the renal artery to cold perfusion. Postoperative parameters for live donors included creatinine levels, hemoglobin levels, and complications at different timepoints. Outcomes of interest were renal function, postoperative complications at various timepoints including 2 weeks and 1 year. Complications were graded using the Clavien–Dindo system. Renal function was assessed by estimated glomerular filtration rate (eGFR), which was calculated from serum creatinine using the modification of diet in renal disease (MDRD) equation. Data on recipients of the 95 donor nephrectomies were also obtained. Recipient eGFR and creatinine at 1 month, 6 months, and 1 year after transplantation were collected along with incidents of delayed graft function and graft failure.

The surgical techniques for RDN and LDN were similar to procedures described in the previous literature [20, 21]. LDN was performed with 3 or 4 ports, depending on the laterality of the kidney, with possible hand assist. In the left LDN, a 5 mm port was placed near the umbilicus with two 12 mm ports at the mid-clavicular line in the lower quadrant and left of the midline in the upper quadrant. For the right LDN, the configuration is similar, but reflected about the midline with an additional 5 mm port at the superior midline for retraction of the liver. RDN was performed with 4 or 5 ports. In the left RDN, a 12 mm port was placed lateral to the umbilicus. Additional 8 mm ports were placed at three locations left of the midline in the upper quadrant, at the mid-clavicular line in the lower quadrant, and at the most lateral portion of the retroperitoneum above the iliac crest. An assistant 12 mm port was placed at the midline, below the umbilicus. The configuration is reflected for the right RDN with the additional of the 5 mm port for liver retraction. In all cases, the kidney was delivered through either a Pfannenstiel incision or a modified Gibson incision.

Patient characteristics and perioperative outcomes were compared between the two groups. Categorical variables were compared using the Chi-squared test. Continuous variables were compared using the Mann–Whitney U test. Statistical significance was set at a p value of 0.05 or less. Statistical analysis was performed using SPSS version 23 (IBM© Corp., Armonk, NY, USA).

Results

Between 2011 and 2016, a total of 95 individuals who underwent either laparoscopic donor nephrectomy or robot-assisted laparoscopic donor nephrectomy performed by a single surgeon were reviewed. Of the 95 patients, 73 patients underwent LDN, while the other 22 patients underwent RDN. Patient demographics are summarized in Table 1. No significant differences in age, sex, BMI, kidney laterality, race, and ASA status were found between the groups. Similarly, there were no significant differences in preoperative serum creatinine, hemoglobin, and eGFR.

Table 1 Demographic and clinical characteristics of donors
Full size table

Intraoperative data are shown in Table 2. RDN was associated with longer operating times (192.3 vs. 149.8 min, p < 0.001). Over the course of the robotic cases, the operating time of RDN eventually approached the mean operating time of LDN with each subsequent procedure. Figure 1 shows the learning curve of RDN operative time vs. surgery date fitted to a cubic regression with the mean operative time of LDN cases as reference. Warm ischemia times (3.4 vs. 2.2 min, p = 0.002) and extraction times (2.6 vs. 1.7 min, p = 0.05) were also found to be longer in the RDN group. There were no significant differences in estimated blood loss or need for blood transfusion. Hospital stay length was similar in both groups (2.5 vs. 2.8 days, p = 0.365).

Table 2 Donor intraoperative outcomes
Full size table
Fig. 1
figure 1

Learning curve of RDN with operative time vs. date of surgery. The learning curve was fitted to a cubic regression (R 2 = 0.286). The dashed line is mean operative time of LDN for reference

Full size image

Postoperatively, decrease in hemoglobin (12.8 vs. 8.3%, p < 0.001) was significantly greater in the RDN group (Table 3). Changes in serum creatinine (26.7 vs. 19.9%, p = 0.086) and eGFR (−22.3 vs. −16.9%, p = 0.086) were found to not significantly differ between the two groups. Postoperatively at 2 weeks, these changes in serum creatinine (47.7 vs. 51.1%, p = 0.481) and eGFR (−35.1 vs. −37.2%, p = 0.481) from baseline did not significantly differ. Similarly, at 1 year follow-ups, changes in serum creatinine and eGFR were not significantly different.

Table 3 Donor postoperative outcomes
Full size table

Table 4 summarizes the postoperative donor complications reported at three different timepoints postoperatively, 2 weeks postoperatively, and 1 year postoperatively. Postoperatively, 4 (5.5%) patients in the LDN group and 2 (9.1%) patients in the RDN group developed complications. Most of the complications were minor complications (Grade I–II), which consisted of fever, nausea, atelectasis, and chylous leakage. The only major complication (Grade IV) in the LDN group was acute pulmonary edema. At 2 week follow-up, there were 11 (15.1%) minor complications in the LDN group and 1 (4.5%) minor complication in the RDN group. Complications in the LDN group included nausea, abdominal/flank pain, epididymitis, seroma, arm paresthesia, wound infection, and diffuse rash. The single minor complication in the RDN group was the development of chylous ascites. At 1 year follow-up, there were 3 (4.1%) minor complications in the LDN group and 1 (4.5%) minor complications in the RDN group. These complications consisted of nausea and testicular soreness/swelling. There were no major complications at 1 year.

Table 4 Donor postoperative complications
Full size table

There was no difference in transplant patient and graft outcome regardless of the donor procedure performed. Transplant recipient renal function did not differ at 1 month, 6 months, or 1 year after transplantation (Table 5). In recipients who received kidneys from donors undergoing LDN, there were three cases of graft failure and one case of delayed graft function. For the three cases, graft failure did not occur immediately, but rather a few years after the transplantation in each case. There were no cases of graft failure or delayed graft function in recipients of donors undergoing RDN.

Table 5 Recipient outcomes
Full size table

Discussion

Live donor nephrectomy is a unique operation performed on healthy donors. As such, it is imperative to maintain the safety of the patient. Although LDN is currently accepted as the gold standard for various reasons, safety, along with its benefits and risks, had to be carefully assessed during its introduction before being widely adopted over open donor nephrectomy. Its reported benefits of shorter hospital stay, less pain, improved cosmesis, and faster return to work were balanced by longer operative times, longer WIT, steeper learning curve, and possible overlooked complications due to limited vision and control [22, 23]. In the same manner, the potential benefits and risks of RDN must also be assessed with regards to safety before it can be considered a viable alternative to LDN.

No differences were found in patient characteristics between RDN and LDN. With regard to perioperative outcomes, we found no differences in blood loss, transfusions, and length of hospital stay when compared to LDN. In this early experience, operative time, warm ischemia time, and extraction time were found to be longer in RDN as well as having greater hemoglobin change. The difference in hemoglobin change was not clinically relevant (−1.2 vs. −1.8 g/dL). The longer times could be a result of our cautious, slower approach with RDN, which was due to our initial unfamiliarity with the procedure. Longer warm ischemia times and extraction times could also be attributed to the extraction being performed by a second attending surgeon, because the primary attending surgeon was at the console, operating the robot. As more robotic cases were accomplished, the surgical technique was further refined and these times improved. In one instance, we found that extensive fat dissection around the kidney allowed for easier and faster retrieval of the kidney. Bhattu et al. also reported longer warm ischemia time and extraction time associated with RDN but similar operating time in both groups (Bhattu 2015). In their case, they explained the longer times as a result of unlocking the fourth arm during extraction of the kidney [24].

As RDN is a relatively new procedure, the learning curve is a possible cause of differences in a number of intraoperative factors including operative time and warm ischemia time. In our case, the surgeon performing the procedures had extensive experience with both laparoscopic and robotic procedures, but not necessarily the specific application of robotic surgery to donor nephrectomy. Because of the surgeon’s extensive experience, we were able to minimize confounders related to operating the DaVinci robot and focus on execution of the procedure with robotic assistance. In our case, operative times were longer with the first few cases, but began to approach the mean LDN operative times by the last case (Fig. 1). Longer WIT could also be due to the learning curve as it has been previously shown that portions of RDN performed under warm ischemia take longer to master than less critical portions of the procedure [25]. We also realize that due to the surgeon’s extensive experience that this learning curve may not be indicative of a surgeon with no prior robotic experience. For newer surgeons without robotic experience, they may face a steeper learning curve due to the challenges of learning both the robotic system and the procedure. Conversely, a surgeon with robotic experience can have a significantly easier time in learning to perform RDN compared to LDN.

From our experience, operative times of robotic cases continued to improve and approach laparoscopic times, similar to other robot-assisted procedures such as robot-assisted partial nephrectomy [26]. In this regard, it is not unreasonable to expect proficiency with RDN to improve with further cases and reach a level of mastery equivalent to that of LDN. With respect to training surgeons in surgical robotics, it has been previously demonstrated that surgeons with formal robotic training are able to perform robot-assisted cases immediately out of training at an equivalent level to experienced surgeons [27]. With regard to mastery, a previous study demonstrated that improvements in robotic-assisted procedures such as partial nephrectomy can still continue to improve up to 300 cases [28]. Our experience with donor nephrectomy shows that mastery may be achieved with significantly fewer cases as we approached standard laparoscopic donor nephrectomy operative times by the 22nd case. However, the pool of donors is limited and will result in a lower number of available donor nephrectomy cases, which may hinder complete mastery. For this reason, simulation may be a crucial tool for training novices. It has been previously reported that it may take 10 h of simulation training for residents to achieve surgical proficiency [29]. Cadavers could be another method of training. However, multidisciplinary cooperation is required for maximal efficiency of the expensive cadaver [30, 31]. The development of a surgical training model may also be useful in overcoming the learning curve for RDN, similar to the training model previously used in robot-assisted partial nephrectomy [32].

Although the operative time, WIT, and extraction time were longer in RDN group, we demonstrate that this does not translate into poorer recipient renal function. Recipient eGFR at 1 month, 6 months, and 1 year post-transplantation was shown to be comparable between the two groups. In addition, there were no cases of delayed graft function or graft failure of recipients receiving kidneys from donors undergoing RDN. A previous study also showed no differences in recipient renal function despite increases in WIT and extraction time [24]. It has also been demonstrated previously that WIT up to 720 s did not correlate with recipient graft function [33]. In all cases, the WIT was well under 720 s, with the maximum WIT in the RDN group being 390 s. We can further expect continued improvement in WIT as we performed more robotic cases after this initial experience. Despite increases in operative times, we also show on the donor side that postoperative renal function was not affected. eGFR was not different between the two groups postoperatively and at 2 weeks and 1 year follow-ups, suggesting that RDN does not negatively affect donor renal function compared to LDN.

The postoperative course, including hospital stay length, did not differ between the two groups. The rate of postoperative complications did not differ postoperatively, at 2 week follow-up and 1 year follow-up. It can also be noted that chylous ascites and leakage were two minor complications that appeared exclusively in the RDN group. These complications prompted further investigation and refinement in our surgical technique which materialized in the form of less fine dissection around the hilum, aorta, and vena cava. With less fine dissection, we were able to lower the likelihood of unintentional damage to the lymphatic vessels. Not only did this prevent future chylous leakage, it also allowed for the elimination of unnecessary actions and faster dissection in later cases. Thus, examination of exclusive complications in robotic cases is crucial for further refinement in the technique.

We acknowledge several limitations of this study. This is a retrospective review of a relatively small sample size of patients performed by a single surgeon. Data, such as cost, cosmesis, and pain, could not be obtained. Although reviewing patients from a single surgeon minimized surgical confounders, the learning curve of RDN shown in this study is from a surgeon with extensive experience in robotic and laparoscopic surgeries, and may not be representative of most surgeons learning RDN. Additional studies on the comparison between RDN and LDN at different stages of the learning curve may be done to resolve this issue. Similarly, the procedures being performed at different timeperiods could be affected by factors such as advancements in knowledge or technology. Although we have compared the longest postoperative courses of patients undergoing RDN and LDN so far, longer term studies may be needed to fully elicit the complete long-term differences between the two procedures.

Despite these limitations, this study has currently the longest postoperative follow-up for both donors and recipients, and demonstrates that RDN is a safe application of robotic surgery with comparable outcomes compared to LDN. Although longer operative time, WIT, and extraction time were initially associated with RDN, substantial improvements were made over the course of this early experience, and further advances can be expected after additional experience with RDN. These differences in time did not have a significant impact on postoperative donor renal function, donor complication rates, and recipient outcomes. Furthermore, the learning curve derived from our experience suggests that a surgeon learning RDN can quickly match LDN operative times.