Robotic vs. laparoscopic major hepatectomy

The introduction of laparoscopic technology and surgical robots in hepatobiliary surgery in the 1990s and 2000s, respectively, has dramatically revolutionized the field. Even though laparoscopic and robotic major hepatectomy was slower to adopt compared to minimally-invasive minor hepatectomy, the number of major hepatectomies performed with both approaches worldwide has significantly increased and is still rising. Despite the few comparative studies between laparoscopic and robotic major hepatectomy, most studies are focused on describing the procedures or reporting the outcomes of each method, either separately, or mixed with minor hepatectomies. Based on the available data, the direct comparison between the two techniques has shown that when robotic major hepatectomy is performed by experienced hepatobiliary surgeons in high-volume centers, it can lead to similar operating times, estimated blood loss, hospital length of stay, complication and mortality rates compared to its laparoscopic counterpart. The likelihood of achieving a margin-negative resection in cancer patients, as well as long-term disease-free and overall-survival are comparable between the groups. However, broader adoption of the robotic approach might be a hurdle in low-volume centers due to the high fixed capital and annual maintenance cost of the surgical robot.


INTRODUCTION
The introduction of minimally-invasive technology in the approach of liver disorders in the early 1990s has since revolutionized the field of liver surgery [1][2][3][4][5] . Laparoscopic liver surgery does not only include pure laparoscopy, but also hand-assisted laparoscopic, as well as hybrid approaches, where the initial part of the procedure (i.e., liver mobilization, early dissection) is done laparoscopically, while later a small incision is made to complete the transection of the liver parenchyma [6,7] . The liver is classified in individual territories according to the segmentation of the vessels and bile ducts, introduced by Couinaud in the 1950s [8,9] , and the Brisbane 2000 nomenclature is utilized to define minor and major hepatectomy in the field of liver surgery [10,11] . Minor hepatectomy is defined as the resection of two or fewer Couinaud segments, while major hepatectomy is the removal of three or more Couinaud segments [11] . The first series on laparoscopic liver resections consisted mostly of minor liver resections [3,4,12,13] . The first laparoscopic major hepatectomy (LMH) was performed in 1997 [14] . The higher risk for uncontrolled hemorrhage and the requirement of advanced technical expertise, particularly related to major vessel dissection, have slowed the broader adoption of minimally-invasive approaches for major hepatectomy [15] .
The technological advances of our era have also led to the broader implementation of robotics in several fields of surgery, including liver surgery. The ability to obtain three-dimensional and magnified intraoperative vision, the significant decrease in hand tremor, as well as the benefit for the surgeon of operating under more relaxed and comfortable circumstances, have led to a considerable growth in robotic surgery, which can overcome the rigid instrumentation and the limited two-dimensional vision associated with laparoscopic surgery [16,17] . These characteristics, along with the advent of wristed instruments, can lead to improved dexterity and higher precision in surgical dissection; this is of particular benefit to liver resection, as hilar dissection, curved transection of the liver parenchyma and the resection of lesions in the posterosuperior segments can be more feasible with the use of a robot [18] . The first large series of robotic liver resection was reported in 2002 [19] , and although most current experience is based on minor resections, several studies have reported robotic major hepatectomy (RMH). This review aims to summarize the current state of evidence about the outcomes after LMH vs. RMH. We acknowledge that there is still a very important role for open hepatectomy in cases of multiple bilobar liver tumors or large tumors near critical vascular structures. However, we will focus on the differences between LMH and RMH, as a full review of open major hepatectomy is beyond the scope of this review.

INTERNATIONAL CONSENSUS AND LEARNING CURVES
Before engaging in a head-to-head comparison between LMH and RMH, it is worth mentioning two points that may favor the former approach. First, LMH has been performed for many more years than its robotic counterpart; second, irrespective of the procedural, hospitalization, and total economic cost, the cost of purchasing a robot for a hospital is considerable and has been a major limiting factor to the broader adoption of robotic liver surgery. These two points are of paramount importance, as data suggest that outcomes improve as experience with a surgical approach grows [20] . It is also worth mentioning that during the second international consensus on laparoscopic liver surgery (Morioka 2014), the jury concluded that laparoscopic minor hepatectomy had at that point already become standard practice, while LMH was still considered to be an innovative procedure still under exploration [11] . According to the 2018 international consensus statement on robotic hepatectomy, RMH was deemed to be as safe and feasible as both LMH and open major hepatectomy [21] .
For the purpose of this review, we performed a non-systematic search of the PubMed bibliographic database using combinations of the following terms: "laparoscopic", "robotic", "minimally invasive", "hepatectomy", "major hepatectomy", "liver resection", and "major liver resection" (last search March 2020). We included comparative or non-comparative studies reporting on the number of LMH and RMH cases. Tables 1, 2, and 3 present the previously published cases of RMH and LMH [6,7,12-14,20,  , and it is apparent that the experience with LMH is greater than that of the robotic approach. Tsung et al. [20] 2014 USA Nov 2007-Dec 2011 57 21 n/a n/a Spampinato et al. [94]  Chen et al. [30,31]  Khan et al. [37] 2018 International 2006-2016 61 16 8 8 Goja et al. [38] 2019 India Feb 2015-Jan 2016 21 6 3 3 Lim et al. [39] 2019* Gravetz et al. [43] 2019 USA 2013-2017 33 8 n/a n/a Magistri et al. [44] 2019 Sucandy et al. [47] 2020 USA 2013-2018 80 24 14 6 Beard et al. [48]  Gumbs et al. [55] 2008 France n/a 3 3 0 0 Gumbs et al. [56] 2008 France n/a 5 5 0 0 Cho et al. [57] 2008 South Korea Jan 2004-Dec 2007 128 47 23 13 Buell et al. [13] 2008 USA Jan 2001-Apr 2008 253 69 24 33 Topal et al. [58] 2008 Dagher et al. [59] 2008 France Since Feb 1999 20 20 0 20 Wakabayashi et al. [60] 2009 Japan Jul 1995-Apr 2008 176 39 10 12 Castaing et al. [ Lim et al. [39]   Determining the learning curve for each approach is also of major significance. The learning curve is "the improvement in performance over time or the change in the ability to complete a task until failure is decreased to a constant acceptable rate" [110] . Data suggest that the learning curve for LMH is around 45-60 cases [93,[111][112][113] . van der Poel et al. [93] reported that 55 is the "golden" number for LMH; however, all surgical operations were performed by two experienced hepatobiliary surgeons with at least three years of additional experience on minor laparoscopic hepatectomy. For RMH, Chen et al. [30] described an initial phase of 15 patients followed by an intermediate phase of 25 patients. The accumulated experience of the first 15 cases (defined as the "initial learning curve"), mostly comprised of right and left hemihepatectomies, was followed by more complex cases, such as trisectionectomy and 8-5-4 trisegmentectomy, in the next 25 cases ("phase of increased competency"). Their last 52-case "matured phase" was associated with an overall improvement in outcomes. However, the authors did not mention who their "learning curve" refers to, as "all procedures were performed by the same operative team", but they do not specify their prior experience with minor robotic resections or even with LMH. Tsung et al. [20] reported that the outcomes of their robotic cases between 2010-2011 were superior to those of the robotic cases between 2007-2010, but the authors pooled together both minor and major resections for this comparison.

OPERATING TIME
A systematic review and pooled analysis of outcomes on robotic liver resections showed that the mean operating time for RMH (≥ 4 segments) was 405 ± 100 min [18] , while another more recent systematic review reported similar pooled mean operating rime for RMH (≥ 3 segments) of 403.4 ± 107.5 min [114] . A systematic literature review on LMH [115] showed that mean operating time in all individuals studies was lower than the pooled operating times reported in the RMH systematic reviews [18,114] . Additionally, in a systematic review comparing LMH to open major hepatectomy, the pooled mean operating time in the LMH arm was 285 ± 105.6 min [116] . Similarly, in a large multicenter study from Europe, Cipriani et al. [109] reported a median operating time of 300 min (IQR 205-380) for LMH, and more specifically 300 min (IQR 240-402) for right hepatectomy and 270 min (IQR 160-290) for left hepatectomy. Tsung et al. [20] compared RMH vs. LMH, and showed that both overall operating room time (452 min vs. 348.5 min) and operating time (330 min vs. 280.5 min) were significantly longer in the RMH group. Spampinato et al. [94] also showed that operating time was longer in RMH (430, IQR 240-725 min) when compared to LMH (360, IQR 180-600 min), while all procedures were performed by surgeons experienced in minimally-invasive liver surgery. Notably, a more recent study showed no difference in median operating time between RMH (194, range 152-255 min) and LMH (204, 149-280 min), and all of the operations were again performed by experienced minimally-invasive hepatobiliary surgeons [42] . A Korean group recently published the initial experience of a single surgeon with robotic liver surgery and showed that there was no difference in operating time between robotic and laparoscopic left hepatectomy (248.6 ± 37.5 min vs. 226.7 ± 26.6 min) [45] . Another recent study comparing robotic vs. laparoscopic right hepatectomy demonstrated that operating time was significantly shorter in the robotic group compared to the laparoscopic one (425 ± 139 min vs. 565.18 ± 183.73 min), and all procedures were performed by the same young surgeon [40] . That may serve as an indicator that as experience with RMH grows, operating time seems to decrease and to be equivalent to, or even shorter than, that of LMH. However, a major confounding factor is surgeon's surgical expertise and prior experience with minimallyinvasive major hepatectomy; thus, future studies comparing operating time, as well as other parameters, between RMH and LMH should always mention primary surgeon's prior experience and should make sure that the two comparison groups are equivalent regarding this parameter.

ESTIMATED BLOOD LOSS
The pooled estimated blood loss (EBL) in RMH based on two systematic reviews was 543.4 ± 371 mL [114] and 380 ± 505 mL [18] , respectively. The pooled mean EBL for the LMH arm in a systematic review comparing LMH to open major hepatectomy was 450.6 ± 563.2 [116] , which is comparable to the pooled rates reported in the RMH systematic reviews [18,114] . However, major deviations were found between the individual RMH or LMH studies themselves included in each systematic review. Cipriani et al. [109] reported a median EBL of 350 mL (IQR 125-1350) for LMH, and more specifically 400 mL (IQR 200-800) for right hepatectomy and 300 mL (IQR 50-260) for left hepatectomy. Studies directly comparing EBL between RMH and LMH showed that EBL in RMH was lower than that in LMH, while the difference was not statistically significant in any of the individual studies [20,40,42,94] .

LENGTH OF STAY
Two prior systematic reviews on RMH reported a pooled mean hospital length of stay (LOS) of 10.5 ± 4.8 [114] and 11 ± 6 days [18] , respectively. The mean LOS of most individual studies included in a systematic review on LMH [115] was shorter than that of the two RMH systematic reviews. Another systematic review showed that the pooled mean LOS for LMH was 10 ± 8.7 days [116] . Cipriani et al. [109] reported a median LOS of 6 days (IQR 4-10) for LMH, and more specifically 7 days (IQR 4-13) for right hepatectomy, and 5 days (IQR 4-10) for left hepatectomy. Studies reporting on the direct comparison of RMH vs. LMH did not demonstrate any statistically significant difference between the two arms [20,40,42,94] .

ECONOMIC COST
Mejia et al. [46] reported that the adjusted room and board charges were significantly lower in the LMH vs. the RMH group, with no other difference between the two groups regarding economic cost. Of note, when comparing the cost of LMH vs. RMH, the fixed capital cost ($1,000,000-$2,600,000 for a robotic system with a 10-year longevity period) [117][118][119][120] and annual maintenance cost ($90,000-$175,000) [120] for a hospital to purchase and maintain a surgical robot, should also be taken into consideration. The addition of this cost can be burdensome, particularly for low-volume liver surgery centers, and this remains a significant driving factor for the slow spread of RMH and robotic liver surgery in general. It should also be noted that access to the robot in the operating room can be a challenge due to competition with other surgical service lines.

CONCLUSION
The introduction of laparoscopy and robotic surgical systems in liver surgery has significantly changed the current state of practice. Although both approaches have been more widely tested for minor liver resections, the number of LMHs and RMHs performed worldwide has significantly increased over recent years, and is still on the rise. Although there is a considerable deviation in outcomes after RMH, especially during early experience, when RMH is performed by experienced surgeons in high-volume liver centers, it can be associated with equivalent operating time, EBL, LOS, morbidity and mortality, and comparable oncologic outcomes in terms of achieving a margin-negative resection and long-term overall survival. The fixed capital and annual maintenance costs for the robotic surgical system may pose a significant obstacle in the broader adoption of RMH, particularly in low-volume centers.