**1. Introduction**

Living kidney donation is the ultimate treatment for end-stage renal disease (ESRD) [1]. Since the first successful living kidney donation in 1955 was carried out by Murray et al., many advances in surgical techniques and immunosuppressive therapy have led to substantial improvements in life expectancy and quality of life, not only for kidney recipients, but also for kidney donors [2]. In particular, minimally-invasive approaches for donor nephrectomy (DN) have increased the incidence of living kidney donation since the first laparoscopic DN (LDN) in 1995 and the first robot-assisted DN (RDN) in 2000 [3–5]. Unfortunately, higher donation rates have not been able to compensate for higher demand, which has led to at least 120,000 patients worldwide waiting for a kidney transplant today.

Many variations of minimally-invasive DN techniques have been described so far. Apart from hand-assisted methods as a bridge to open surgery, DN has also been performed in a retroperitoneoscopic (hand-assisted) manner [6,7]. In line with shorter flank incisions for open DN ("minimally invasive" open DN), Gill et al. conducted the first LDN via a LESS approach (laparoendoscopic single site surgery) in 2008 and inserted all trocars through the umbilicus [8,9]. Others have even tried to perform DN as a NOTES (natural orifice transluminal endoscopic surgery), and Pietrabissa et al. were the first to report a transvaginal extraction of the kidney after RDN in 2010 [10]. Today, some high-volume centers have performed more than 100 RDNs or LESS single-port RDNs, and employ specialized robotic single-site platforms [11,12]. However, the robotic approach still accounts for less than 5% of all minimally-invasive DNs, with increasing incidence compared to conventional transperitoneal LDN at more than 50% [13].

Irrespective of this magnitude of variations, minimally-invasive approaches for donor nephrectomy represent the standard of care, and are recommended as "the preferential technique", according to the current guidelines for renal transplantation of the European Association of Urology (EAU) [14,15]. Multiple studies have shown that LDN is superior to open DN (ODN) in terms of hospital stay or postoperative pain, but the operating and warm ischemia time (WIT) are longer [16]. Importantly, LDN is not inferior in terms of complication rates, short- and long-term graft function. On the other hand, when comparing LDN with the robotic approach, RDN appears to have even less postoperative pain and less blood loss, but a longer WIT and operating time [17]. Nonetheless, analyses of cohorts with big sample sizes are still lacking, and the high variability of minimally-invasive DN renders it di fficult to draw direct conclusions.

With this in mind, we conducted a retrospective bi-centric comparison of transperitoneal LDN with RDN and included more than 250 interventions. We aimed to compare perioperative outcomes as well as short- and mid-term kidney function of the donor and recipient up to four years after surgery. Alongside sub-analyses controlling for inherent learning, regression analyses to predict postoperative kidney and graft function were performed. All LDNs were conducted at the largest German kidney transplant program run by a urologic department that has been performing LDNs since 1999. All RDNs including the very first RDN in Germany in 2007, were performed at a urologic department highly specialized in robotic surgery [18].

#### **2. Materials and Methods**

In total, 257 DNs performed at two tertiary referral centers were retrospectively analyzed. All 205 LDNs were conducted by 11 surgeons with a median caseload of 11 (range 2–43) at a urologic department specialized in laparoscopic kidney surgery including LDNs. The 52 RDNs were performed at another urologic department, which is specialized in robotic surgery in general. All RDNs were conducted by five surgeons with a median caseload of 10 (range 2–29). The interventions were performed in a transperitoneal fashion between 2007–2020 (RDN) and 2011–2016 (LDN).

At the robotic department, the very first RDN in Germany was conducted [18]. Before 2007, all donor nephrectomies had been held in an open fashion, so none of the robotic surgeons had prior expertise in LDN, but in a large variety of other robotic interventions. Thereafter, DN was standardized to a robot-assisted approach. The other department in this study has been performing LDNs since 1999. Both departments always conducted DNs in a minimally-invasive fashion during the study period, unless the donor had a significant amount of prior abdominal surgeries and consequently high risk for conversion. The corresponding kidney transplantations were held in an open fashion, except for the last 18 (34.6%) cases at the robotic department. As a part of the EAU-RAKT working group (European Association of Urology working group for robotic kidney transplantation), the first RAKT in Germany was performed there in June 2016 [19,20]. From then, all RDNs were followed by RAKTs.

This entire analysis was conducted in adherence with the correct scientific research work terms of the Charité Medical University of Berlin and Saarland University including full anonymization of patient data. All the patients included in the analysis provided written informed consent.

#### *2.1. Surgical Technique*

All RDNs were performed using a transperitoneal approach, with either a DaVinci® Si or X system with four arms. The ports were placed pararectally. For the first RDNs, the graft was removed in a hand-assisted manner without a specimen bag via a Pfannenstiel incision, and later on via a periumbilically placed GelPOINT® trocar (Applied Medical, Los Angeles, CA, USA). For LDN, the approach was purely laparoscopic, without the hand-assisted technique, which has been described previously [21,22]. In brief, four ports were used, and the kidney was extracted through an enlarged lateral trocar incision measuring 5 to 6 cm.

#### *2.2. Data Collection and Outcome Measures*

For the donor characteristics, age, gender, body mass index (BMI, kg/m2), pre-existing arterial hypertension, diabetes, and history of smoking were obtained. The graft's side, scintigraphic split-renal function (DTPA), and number of arteries and veins served as organ-specific factors. For the recipient characteristics, age, gender, BMI, implantation side, and individual number of prior kidney transplantations were obtained.

Intraoperative (operating time, WIT, complications) and postoperative (length of stay, major postoperative complications based on Clavien–Dindo grade≥3 within 30 days after surgery) results were analyzed as *primary outcomes*. The comparison and prediction of postoperative kidney function of the donor and of the recipient up to four years after transplantation served as *secondary outcomes*. Delayed graft function (DGF), defined as dialysis within one week after transplantation or insufficient serum creatinine decline not below 2 mg/dL, was analyzed as a further kidney-related secondary outcome.

#### *2.3. Statistical Analysis*

Primary and secondary outcomes were compared between the LDN and RDN group. To assess whether perioperative outcome was affected by an inherent learning curve, both groups were split in half and the outcomes were compared within each group. The first 34 (65.4%) RDNs were followed by an open transplantation, but the last 18 (34.6%) were followed by a robot-assisted kidney transplantation. To ensure that RAKT did not affect the perioperative results of RDN, the last 18 RDNs were excluded in another sub-analysis. The impact of patient-, graft- or surgery-specific factors on postoperative kidney function of the donor at discharge was assessed by linear regression analysis. To predict kidney function of the recipient one week after surgery, donor and recipient characteristics, DN, and transplantation-specific aspects were included in another uni- and multivariate regression analysis.

Categorical variables were reported as frequencies and proportions, and continuous data as the median and range. Fisher's exact test and the Mann–Whitney U test were used to compare between groups. Covariates were included in the multiple regression analysis only if their respective effect was significant in the univariate analysis. The statistical analysis was performed by SPSS version 25 with Fix pack 2 installed (IBM, Armonk, NY, USA). All tests were two-sided, and *p*-values < 0.05 were considered significant.
