Robot-assisted vs traditional laparoscopic partial nephrectomy: the time for meta-analysis has not yet arrived


International guidelines recommend partial nephrectomy (PN) in all cT1 cases, whenever technically feasible. Regardless of the approach used, PN must be finalized to achieve the complete removal of the tumour, whilst preserving the largest possible part of healthy renal parenchyma and avoiding major early and/or late complications. Obviously, the feasibility of PN can be influenced by tumour-, patient- and surgeon-related factors; anatomical and topographical tumour characteristics , as well as patient comorbidities and body mass index, should be carefully considered so as to identify the cases suitable for PN. In addition, the most important factor in the decision-making process can be the surgeon's experience with a specific approach and/or technique .

For many years open surgery represented the ‘gold standard’ approach for PN and it is possible that the majority of surgeons remain more confident in treating more complex tumours using this classic approach; however, open surgery requires a non-aesthetic skin incision and painful access through the muscular plane and also entails a long hospital stay and postoperative recovery time. In addition, the risks of chronic pain, herniation or muscle relaxation at the level of the lumbar or abdominal region are underestimated problems that can limit patient satisfaction after this approach. In recent years, traditional laparoscopic PN (LPN) and robot-assisted laparoscopic PN (RAPN) have been recommended as these techniques minimize the invasiveness of open surgery, offering patients better cosmetic results, less postoperative pain, and shorter hospital stay and postoperative recovery time.

To date, LPN has been considered to be a challenging procedure requiring a high volume of cases to complete the steep learning curve in an acceptable time and to achieve the optimum level of skills required. Instruments with a limited degree-of-freedom make it technically difficult to follow favourable angles for tumour excision, haemostasis, repair of the collecting system and reconstruction of the parenchymal defect. The consequence was the limited diffusion of this technique into the hands of few very expert surgeons able to apply the procedure for the treatment of more complex cases. The majority of other surgeons continued to prefer the open approach, eventually limiting the laparoscopic approach to technically simpler cases.

Robot-assisted PN seems to be a promising procedure, able to bridge the technical difficulties of LPN and leading to a broader diffusion of the laparoscopic treatment of renal masses. Three-dimensional (3D) vision, optical magnification and the patented EndoWrist instruments (Intuitive Surgical, Sunnyvale, CA, USA) allow surgeons to achieve very precise tumour resection and to simplify the reconstructive steps of the procedure, minimizing warm ischaemia time (WIT) and potential damage to the kidney parenchyma even in more complex tumours. Recently, several single- and multicentre, observational, non-comparative series have reported good perioperative (WIT, operating time and perioperative complications), functional (renal function preservation) and early oncological (positive surgical margin rate) outcomes, which support the use of this approach instead of LPN and as the principal alternative to open PN [1]. The main factor limiting the further diffusion of RAPN, especially in Europe, is now the availability of the da Vinci robot and its relative costs.

Obviously, the available observational, non-comparative studies with limited follow-up cannot confer to RAPN a high level of evidence but, recently, some meta-analyses of available studies comparing LPN and RAPN have been published [2, 3]. After a systematic review of the literature, Aboumarzouk et al. [2] performed a meta-analysis of seven comparative studies and concluded that RAPN appears to be a feasible and safe alternative to its laparoscopic counterpart with decreased WIT. In the current issue of the BJUI, Froghi et al. [3] present another systematic review and meta-analysis of six studies comparing LPN and RAPN. They conclude that there were no significant differences in peri-operative variables between the two approaches. Although both studies were well conducted from a methodological point of view, the different search strategies, inclusion/exclusion criteria for study selection, data extraction of variables of interest and statistical methods used could explain why only one of the selected comparative studies was included in both meta-analyses. Obviously, methodological considerations could also explain the difference in the reported results in terms of WIT. Conversely, the meta-analysis conducted by Froghi et al. confirmed the results previously reported by Aboumarzouk et al., showing that all other peri-operative outcomes, including operating time, estimated blood loss, length of hospital stay and perioperative complications, resulted in overlapping between the two techniques [2]. According to these available meta-analyses, it seems that RAPN does not offer significant advantages in terms of peri-operative outcomes compared with LPN, with a lower WIT being the only exception. Is this really the case? Both meta-analyses provided a combination of data from small, observational studies in which the control group was represented by a historical series (level of evidence 4). In many instances, LPN was performed by expert surgeons who had completed the learning curve. Conversely, RAPNs were usually performed by urologists at the beginning of their experience; therefore, both meta-analyses are representative of a cohort of patients treated during the learning curve of RAPN and this demonstrates clearly that during this step of development equal peri-operative outcomes can already be predicted for RAPN in comparison with LPN performed by more experienced laparoscopic surgeons or by surgeons who have completed the learning curve.

The relevant and negative impact of the learning curve on peri-operative outcomes is further confirmed by the results reported in two other comparative studies not included in the pooled analyses and enrolling >100 cases in the RAPN arm. Benway et al. [4] compared 118 consecutive LPNs with 129 consecutive RAPNs performed by three experienced surgeons at three academic centres in the USA. In this multicentre study, the authors demonstrated a significant advantage in favour of RAPN in terms of WIT, estimated blood loss and hospital stay. More recently, Mullins et al. [5] compared 102 LPNs with 105 contemporary RAPNs performed by the same surgeon. The authors confirmed a significant advantage in favour of RAPN in terms of operating time, WIT, estimated blood loss and risk of major complications. The exclusion for methodological reasons of these comparative studies from both meta-analyses is a relevant concern. Indeed, in both comparative studies, it is possible that the effect of the learning curve was minimized by the number of cases enrolled in the RAPN arm. Moreover, the effect of the learning curve in the pooled comparative studies could explain why peri-operative data reported in the forest plots are not representative of peri-operative outcomes reported in the most recent studies published in the literature [1].

Another point to consider is that most of the available studies comparing LPN and RAPN were performed at an inappropriate stage of the evaluation of this new approach. According to IDEAL recommendations proposed in 2009 by members of the Balliol Collaboration [6], the initial steps to perform an appropriate evaluation of a new surgical technology are 1) an ‘innovation step’ in which the surgical innovation is described for the first time; 2) a ‘development step’ in which the surgical innovation is evaluated in a small group of patients to refine or modify the precise technique; 3) an ‘exploration step’ in which the safety and effectiveness of the surgical innovation can be tested in a context of prospective, single-centre, case-series studies including a large number of cases. Only after these three steps, can the effectiveness of surgical innovation be assessed (the assessment step) against current standard treatment in the context of randomized controlled trials, or using alternative, non-randomized, well-designed comparative studies.

It seems logical that only comparative studies enrolling patients treated in a contemporary period and by expert surgeons who have completed the learning curve should be taken into consideration to really assess the advantages and disadvantages of RAPN against LPN and/or open PN. As long as these studies are not available, pooled analyses must be considered inappropriate and they can only offer wrong arguments to the conservative surgeons and hospital managers, leading them to continue their resistance to new technologies or their drive to limit costs.

In conclusion, the available studies on RAPN show that this approach has completed the first three stages of IDEAL recommendations positively and started the fourth stage (assessment stage) with the usual methodological limits when surgical techniques are compared. Single- and multicentre series have confirmed that RAPN is a safe and effective procedure for the treatment of small, localized renal tumours. Preliminary studies comparing an appropriate number of cases showed significant advantages in terms of peri-operative outcomes in comparison with LPN.

In the years to come, the urological community must favour comparative studies enrolling patients treated in the same period (better if prospective), by expert surgeons who have completed the learning curve and who use a similar surgical technique. Moreover, the different PN approaches will have to be compared, appropriately stratifying the outcomes according to the different anatomical and topographical tumour characteristics. Only when these (randomized or non-randomized) well-conducted studies are available, can the systematic reviews of the literature be enriched by pooled analyses evaluating the most relevant peri-operative, functional and oncological outcomes. In parallel to this, we must wait for an appropriate follow-up so as to complete the last stage of the IDEAL evaluation which is the long-term evaluation of oncological and functional outcomes.