**4. Discussion**

In this bi-centric study, a comparison of 257 minimally-invasive donor nephrectomies with 205 laparoscopic and 52 robot-assisted DNs was conducted. Of note, this analysis included the very first RDN in Germany, and all LDNs were performed at a urologic department where LDNs have been conducted since 1999 [18].

Concerning the primary outcomes, operating time was no different between RDN and LDN (223.5 vs. 213 min, see Table 1). Most studies describe shorter operating times for LDNs, but report highly variable results [17]. Mean operating times for RDNs range from 144 to 306 min [23,24], and for LDNs between 178 and 270 min [25,26], even when only studies with cohorts larger than 100 patients are included. These differences could result from inherent learning curves: Horgan et al. and Janki et al. have shown that operating times in RDN shorten with growing expertise [27,28]. Interestingly, our data do not show an inherent learning effect, either in the RDN or in the LDN cohort. Outcomes remained unchanged over time (see Table 3). Conversely, operating time became significantly longer within the second half of the RDNs (185 vs. 265 min, *p* < 0.001).

This counterintuitive development resulted from the way transplantations were organized, as both institutions perform DNs and transplantations in different operating rooms simultaneously, but not sequentially. Two surgical teams work in parallel, but the graft is not removed unless the transplantation team is ready, to avoid long cold ischemia times. The RDN cohort not only comprised the first RDN, but also the first robot-assisted kidney transplantation in Germany (procedure #35) [18,20]. Operating times in the RDN cohort became longer from that point, as the learning curve for RAKTs had not ye<sup>t</sup> been passed. Naturally, the RDN team started more than 30 min before the transplantation team, but RAKT proved to be much more challenging and time-consuming. When excluding the last 18 cases, when RDN was followed by RAKT, the operating times of the RDNs did not change over time. Thus, the obvious lack of a typical learning curve illustrates that for LDNs, the learning curve had already been passed and for RDNs, significant prior expertise in robotic surgery made it possible to reach stable results from the start [29].

As with the operating time, WIT was not different between RDNs and LDNs (3 vs. 2.45 min). In the RDNs, most grafts were extracted via a GelPOINT® trocar (Applied Medical, Los Angeles, CA, USA), which is an easy and fast, ye<sup>t</sup> expensive method. Wang et al. illustrated significantly longer WIT for RDNs than LDNs in their meta-analysis, which is an often-stated argumen<sup>t</sup> against RDNs [17,30]. However, it is unlikely that differences of 30 or 60 s in WIT will harm the graft function in the long-, mid- or even short-term. It has clearly been shown that a WIT longer than 45 min impairs graft survival in living kidney donation [31]. Fortunately, neither our results nor those from other studies have documented WIT longer than 15 min for RDNs, keeping in mind that the consecutive CIT is again followed by another WIT during transplantation.

Intraoperative complication rates were low in both RDNs (5.7%) and LDNs (2.9%), and did not significantly differ. In line with others, most intraoperative complications were bleedings, whereof one in the LDN group made a conversion to open surgery necessary, but none in the RDN group [17]. In contrast, a patient with massive obesity had multiple trocar dislocations within the first minutes of surgery, so the RDN had to be converted to open surgery. Due to a technical defect of the stapler system for one patient in the RDN group, which made it cut but not staple, locking Hem-o-Lok clips were predominantly used later on, as described elsewhere [32]. During LDNs, Hem-o-Lok and titanium clips are used for the renal artery, a stapler for the right vein, and two Hem-o-Lok clips for the left vein. Not only intraoperative but also postoperative complication rates, according to Clavien–Dindo, were low and did not differ between LDN and RDN. Therefore, both surgical approaches had equivalent complication rates, while LDN has less costs, but RDN appears to be superior in complex situations such as bleedings.

The kidney donors were discharged five days after DN, irrespective of the type of surgery (see Table 2). Consequently, the median length of stay was longer than in most other works, ranging from 2–3 days for LDNs and RDNs [11,17,24]. This can be attributed to di fferences in national health care systems as (i) the German reimbursement system covers a longer hospital stay and (ii) most donors wanted to stay longer as inpatients for psychological reasons. In fact, only 15 (5.8%) patients were discharged two or three days after DN. Early discharge after RDN and LDN is possible from a surgical point of view, however, it has not been a crucial parameter for our perioperative approach, as long as neither patient satisfaction nor health care costs are a ffected.

As a secondary outcome, the impact of the surgical approach on postoperative kidney function was assessed. Kidney donors had a worse kidney function at discharge, which was comparable between groups and similar to results found in other studies (RDN 1.1 mg/dL vs. LDN 1.23 mg/dL; see Figure 1) [28,33]. Correspondingly, the preoperative kidney function, but not the type of surgical approach for DN, was predictive for the postoperative kidney function of the donor at discharge (see Table 4). Interestingly, patient gender also had a significant impact on postoperative kidney function. However, this should not be over-interpreted, as male kidney donors had a worse kidney function than women, with higher serum creatinine values preoperatively (0.9 vs. 0.72 mg/dL, *p* < 0.001) and postoperatively (1.42 vs. 1.1 mg/dL, *p* < 0.001) in this analysis. For this reason, (male) patient gender was predictive for (worse) postoperative kidney function; this may not be representative for other cohorts.

Similarly, Benoit et al. created a model to predict 1-year postoperative renal function of kidney donors after LDN, which has been externally validated [34,35]. The authors predicted postoperative eGFR by preoperative eGFR and patient age (postoperative eGFR = 31.71 + (0.5 × preoperative eGFR) − 0.314 × age at donation). In our model, patient age was not predictive for postoperative kidney function, potentially because we evaluated the short-term kidney function at discharge and not one year after DN.

Concerning recipients, the DGF rates of 6.3% (LDN) and 11.5% (RDN) did not significantly di ffer between groups. In general, there is a large variety of reported DGF rates in living kidney donation, ranging from 4 to 10% [36,37]. This not only results from center-specific di fferences, but also from inconsistent definitions: DGF can be defined by urine output per day, serum creatinine decline, or the need for dialysis after transplantation [36]. We applied a considerably broad definition for DGF (postoperative dialysis within one week after transplantation for any cause or insu fficient creatinine decrease not below 2 mg/dL). DGF rates in the RDN group were 11.5% due to transplantation-related surgical, mainly vascular causes. One (1.9%) patient with a mechanic aortic valve developed a perirenal hematoma, causing prolonged creatinine decline without the need for dialysis. In the LDN group, DGF was mainly caused by acute rejections (3.4%), and also comprised one patient (0.5%) who required dialysis for potassium depletion only. Consequently, DGF did not result from the type of DN, but rather transplantation-specific causes.

Regardless, the kidney function of the recipients significantly improved after transplantation, and did not di ffer between groups during follow-up (see Figure 1). In the multiple regression analysis, not only the preoperative kidney function of the recipient, but also a history of donor smoking, had a significant impact on graft function one week after transplantation (see Table 4). Smoking is a well-known modifiable risk factor for the development of chronic and end-stage kidney disease [38,39]. A history of donor smoking has a negative impact not only on the survival of the donor, but also of the recipient [40]. In our cohort, a positive history of donor smoking increased serum creatinine one week after transplantation by 0.63 mg/dL. This highlights, again, the importance of informing not only transplant patients, but also potential kidney donors, about the risks of tobacco use, and the importance of helping patients to stop smoking.

This analysis is not devoid of limitations. As a bi-centric study, experienced but di fferent surgeons and di fferent teams conducted the RDNs and LDNs. Patient cohorts did not significantly di ffer in terms of characteristics, but were not equally balanced in terms of caseload. Although surgical results were not a ffected by inherent learning curves, at least the results in the RDN group were a ffected by simultaneous robot-assisted kidney transplantation. This procedural aspect highlights the complexity of comparing minimally-invasive donor nephrectomies: the surgical part itself is in high demand, but the high variability of the technical, procedural, and underlying ethical aspects also have to be taken into account [41].
