The Safety and Efficacy of Approaches to Liver Resection: A Meta-Analysis

Abstract

Background:

The aim of this study is to compare the safety and efficacy of conventional laparotomy with those of robotic and laparoscopic approaches to hepatectomy.

Database:

Independent reviewers conducted a systematic review of publications in PubMed and Embase, with searches limited to comparative articles of laparoscopic hepatectomy with either conventional or robotic liver approaches. Outcomes included total operative time, estimated blood loss, length of hospitalization, resection margins, postoperative complications, perioperative mortality rates, and cost measures. Outcome comparisons were calculated using random-effects models to pool estimates of mean net differences or of the relative risk between group outcomes. Forty-nine articles, representing 3702 patients, comprise this analysis: 1901 (51.35%) underwent a laparoscopic approach, 1741 (47.03%) underwent an open approach, and 60 (1.62%) underwent a robotic approach. There was no difference in total operative times, surgical margins, or perioperative mortality rates among groups. Across all outcome measures, laparoscopic and robotic approaches showed no difference. As compared with the minimally invasive groups, patients undergoing laparotomy had a greater estimated blood loss (pooled mean net change, 152.0 mL; 95% confidence interval, 103.3–200.8 mL), a longer length of hospital stay (pooled mean difference, 2.22 days; 95% confidence interval, 1.78–2.66 days), and a higher total complication rate (odds ratio, 0.5; 95% confidence interval, 0.42–0.57).

Conclusion:

Minimally invasive approaches to liver resection are as safe as conventional laparotomy, affording less estimated blood loss, shorter lengths of hospitalization, lower perioperative complication rates, and equitable oncologic integrity and postoperative mortality rates. There was no proven advantage of robotic approaches compared with laparoscopic approaches.

INTRODUCTION

Historically, because of visibility issues and the complicated relationship between the liver and its vasculature, hepatectomy has presented a challenge to the surgeon. Laparoscopy for liver resection was first documented in the early 1990s, proving to be as safe as conventional open hepatectomy while retaining oncologic integrity.^1–7 However, laparoscopy has limitations in transection, mobilization, and the ability to control bleeding. To overcome some of these shortcomings, robot-assisted approaches have been devised and implemented that broaden visualization from 2 dimensions to 3 dimensions and increase range of motion to 360° via the EndoWrist (Intuitive Surgical Inc., Sunnyvale, California). Although minimally invasive approaches to surgery are known to decrease postoperative pain scores and length of hospitalization (LOH), with the rising costs of health care, controversy continues to surround discussions of these approaches.

The focus of this study is to evaluate the role of minimally invasive techniques in liver surgery as compared with a conventional open approach. We compared data related to operative time, perioperative complications, LOH, surgical margins, mortality rates, and cost analysis to assess differing approaches. This is the first systematic review to include analysis of the robotic approach, reflecting trends in modern surgery.

MATERIALS AND METHODS

Identification of Trials and Data Extraction

Two independent reviewers conducted a systematic search of PubMed and Embase on articles published until August 2013. The following medical subject headings were used to locate articles: liver robotic, hepatic robotic, hepatic laparoscopic, hepatectomy laparoscopic, and hepatectomy open. The inclusion criteria for articles were as follows: (1) articles comparing conventional open liver resection with either a laparoscopic or robotic approach; (2) controlled clinical trials, multicenter studies, or randomized controlled trials; (3) studies that reported outcomes of intraoperative and postoperative outcomes, including total operative time, estimated blood loss (EBL), LOH, surgical margins, postoperative complications, and postoperative mortality rates; and (4) studies that reported a measure of variance (standard error, standard deviation, or confidence interval [CI]). The references of articles included in the analysis were manually searched for additional articles for inclusion. Excluded from analysis were articles on resection of colorectal cancer with synchronous liver metastasectomy and articles not published in English. In instances in which research groups reported findings using shared patient populations, the earliest publication by that research group was included for analysis. The results from the 2 independent reviews were compared for accuracy, with disagreement resolved by consensus. To achieve completeness and to assemble the most representative patient database, series with limited sample sizes were included so that their experience would find meaning in aggregate.

Statistical Analysis

The primary outcomes of interest in this study were total operative time, EBL, LOH, surgical margins, perioperative complications, and postoperative mortality rates. A cost analysis was included as a secondary outcome of interest. For continuous outcomes, mean net changes were calculated as primary outcomes, whereas for categorical outcomes, odds ratios (ORs) were calculated to examine the treatment effect. DerSimonian and Laird random-effects models were used to pool mean net changes or ORs across the studies.⁸ The presence of heterogeneity was assessed with the Cochran Q test, and the extent of heterogeneity was quantified with the I² index. To assess publication bias, funnel plots were constructed for each outcome. The Begg rank correlation test was used to examine the asymmetry of the funnel plot, and the Egger weighted linear regression test was used to examine the association between the mean effect estimate and its variance. In addition, sensitivity analyses were conducted by excluding each study in turn to evaluate its relative influence on the pooled estimates. All analyses were conducted using Stata software, version 10 (StataCorp, College Station, Texas).

RESULTS

Search Results

Eight hundred seventy-three abstracts were identified, 867 of which were obtained via searches of 2 databases, with an additional 6 retrieved through manual searches of references. The final set of articles undergoing analysis was attained using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (Figure 1).⁹ Of the 873 abstracts identified, 55 underwent full-text review and 49 articles are included in this meta-analysis, with 3 comparing laparoscopic liver resection with robotic hepatectomy^10–12 and 46 comparing laparoscopic liver resection with a conventional open approach (Table 1).^13–58

Figure 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flowchart showing literature search and study selection.⁹ LH = laparoscopic hepatectomy; OH = open hepatectomy; RH = robotic hepatectomy.

Table 1.

Studies Selected for Meta-Analysis

Authors	Year	Country	Journal	Comparison	n
Packiam et al10	2012	USA	J Gastrointest Surg	LHa vs RHa	29
Berber et al11	2010	USA	HPB	LH vs RH	32
Troisi et al12	2013	Belgium	Int J Med Robot	LH vs RH	263
Inoue et al13	2013	Japan	Am Surg	LH vs OHa	47
Slakey et al14	2013	USA	JSLS	LH vs OH	62
Kim et al15	2011	South Korea	J Korean Surg Soc	LH vs OH	55
Abu Hilal et al16	2008	UK	Eur J Surg Oncol	LH vs OH	44
Endo et al17	2009	USA	Surg Laparosc Endosc Percutan Tech	LH vs OH	21
Cai et al18	2009	Germany	Surg Endosc	LH vs OH	38
Ito et al19	2009	USA	J Gastrointest Surg	LH vs OH	130
Morino et al20	2003	USA	Surg Endosc	LH vs OH	60
Belli et al21	2007	Italy	Surg Endosc	LH vs OH	46
Aldrighetti et al22	2008	USA	J Gastrointest Surg	LH vs OH	40
Topal et al23	2008	USA	Surg Endosc	LH vs OH	152
Kandil et al24	2012	USA	Surgery	LH vs OH	36
Cannon et al25	2012	USA	Surgery	LH vs OH	175
Polat26	2012	Turkey	Surg Laparosc Endosc Percutan Tech	LH vs OH	19
Johnson et al27	2012	USA	J Am Coll Surg	LH vs OH	212
Bhojani et al28	2012	Canada	J Am Coll Surg	LH vs OH	171
Tranchart et al29	2010	France	Surg Endosc	LH vs OH	84
Tang et al30	2005	Hong Kong	Surg Endosc	LH vs OH	17
Lesurtel et al31	2003	France	J Am Coll Surg	LH vs OH	38
Cheung et al32	2013	Hong Kong	Ann Surg	LH vs OH	60
Kobayashi et al33	2013	Japan	Surg Endosc	LH vs OH	83
Slim et al34	2012	Italy	Langenbecks Arch Surg	LH vs OH	92
Hu et al35	2012	China	Surg Laparosc Endosc Percutan Tech	LH vs OH	26
Hu et al36	2011	China	World J Gastroenterol	LH vs OH	60
Gustafson et al37	2012	USA	Surg Endosc	LH vs OH	76
Nguyen et al38	2011	USA	Arch Surg	LH vs OH	86
Tu et al39	2011	China	World J Gastroenterol	LH vs OH	31
Vanounou et al40	2010	Canada	Ann Surg Oncol	LH vs OH	73
Castaing et al41	2009	France	Ann Surg	LH vs OH	120
Carswell et al42	2009	UK	BMC Surg	LH vs OH	20
Dagher et al43	2009	France	Am J Surg	LH vs OH	72
Rowe et al44	2009	Canada	Surg Endosc	LH vs OH	30
Sarpel et al45	2009	USA	Ann Surg Oncol	LH vs OH	76
Tsinberg et al46	2009	USA	Surg Endosc	LH vs OH	74
Cai et al47	2008	China	Surg Endosc	LH vs OH	62
Lee et al48	2007	Hong Kong	Hong Kong Med J	LH vs OH	50
Mala et al49	2002	Norway	Surg Endosc	LH vs OH	27
Rau et al50	1998	Germany	Hepatogastroenterology	LH vs OH	34
Shimada et al51	2001	Japan	Surg Endosc	LH vs OH	55
Farges et al52	2002	France	J Hepatobiliary Pancreat Surg	LH vs OH	42
Laurent et al53	2003	France	Arch Surg	LH vs OH	27
Kaneko et al54	2005	Japan	Am J Surg	LH vs OH	58
Polignano et al55	2008	UK	Surg Endosc	LH vs OH	50
Lai et al56	2009	China	Arch Surg	LH vs OH	58
Truant et al57	2011	France	Surg Endosc	LH vs OH	89
Koffron et al58	2007	USA	Ann Surg	LH vs OH	400

^aLH = laparoscopic hepatectomy; OH = open hepatectomy; RH = robotic hepatectomy.

Description of Included Trials and Demographic Data

The 49 articles analyzed represent a total of 3702 patients, with sample sizes ranging from 17 to 400 patients. The distribution of the patients was as follows: 60 in the robotic group, 1901 in the laparoscopic group, and 1741 in the open group. Baseline patient demographic data, including sex, age, and body mass index, were well matched among groups (Table 2) Distribution of resection type by the 40 articles mentioning this characteristic is listed in Table 3.^{11–17,19–25,27–32,34,35,37–39,41–44,46–48,51–58}

Table 2.

Demographic Characteristics

Characteristic	Total (%)	LHa	OHa	RHa
Sex
Male	1535 (48.7)	712	786	37
Female	1329 (42.1)	705	601	23
Age, y	58.95	58.79	58.87	62.73
BMIa	26.67	26.46	26.31	31.00
Lesions
Mean number	3.26	4.35	2.72	1.49
Mean size, cm	5.11	4.86	4.06	27.50
Surgical indication
CRCa metastases	836	396	412	28
Adenoma	117	110	7	0
FNHa	127	109	18	0
Hemangioma	114	85	23	6
HCCa	951	436	509	6
Hydatid cyst	108	86	18	4
Living donor	52	32	20	0
Cholangiocarcinoma	15	8	6	1

^aBMI = body mass index; CRC = colorectal cancer; FNH = focal nodular hyperplasia; HCC = hepatocellular carcinoma; LH = laparoscopic hepatectomy; OH = open hepatectomy; RH = robotic hepatectomy.

Table 3.

Resection Type

Resection Type	LHa	OHa	RHa
Monosegmentectomy	304	250	7
Subsegmentectomy/wedge	270	249	15
Bisegmentectomy	173	141	8
Left lateral sectionectomy	323	231	2
Right trisegmentectomy	7	6	0
Mixed segments	26	0	8
Right hepatectomy	173	169	0
Left hepatectomy	113	72	0
R extended hepatectomy	12	22	0
Major hepatectomy	110	119	0
Nonanatomical/atypical	105	88	0
P-S segment	110	37	22

^aLH = laparoscopic hepatectomy; OH = open hepatectomy; RH = robotic hepatectomy.

Perioperative Outcomes

Forty-six publications reported total operative length, with similar results among groups^{10–13,15–24,26–36,38–58} (Figure 2a). The mean total operative time was 203.6 minutes, 203.9 minutes, and 234.8 minutes for the laparoscopic, open, and robotic groups, respectively.

Figure 2.

Forest plots and pooled analyses of mean difference in length of hospitalization (a), operative time (b), estimated blood loss (c), and odds ratio of postoperative complications (d). The mean difference is reported for each study (black boxes). LH = laparoscopic hepatectomy; OH = open hepatectomy.

Regarding EBL, 44 studies reported this variable.^{10–14,16–25,27–39,43,44,46–58} There was no difference between minimally invasive approaches, and there was a statistically significant increase in blood loss in laparotomy cases as compared with laparoscopy cases, with a pooled net mean change of 152.0 mL (95% CI, 103.3–200.8 mL) (Figure 2b).

The total number of conversions in the laparoscopic group was 106, which represents a 5.68% conversion rate to open surgery. In the robotic group, 9 cases required conversion to open surgery, representing a 15% conversion rate.

Twenty-nine studies included results of pathologic resection margin status in their analyses.^{13–16,19–21,23–25,28,29,31–33,38,41–43,45–49,51,53,56,57} Laparoscopy showed a significantly higher rate of negative surgical margins (pooled OR 1.06) as compared with laparotomy (pooled OR 1.01).

Postoperative Considerations

Forty-four studies reported LOH.^{10,12,13,15–32,34–40,42–44,46–58} As compared with patients undergoing the laparoscopic approach, those undergoing a conventional open approach had a significantly longer LOH (pooled mean difference, 2.22 days; 95% CI, 1.78–2.66 days) (Figure 2c).

Postoperative morbidity, including wound infection, biliary leakage, pleural effusion, bleeding, fluid collection, incisional hernia formation, renal failure, and ascites or cirrhotic decompensation, was reported by 47 articles.^{10–32,34–41,43–58} For total postoperative complications, minimally invasive approaches showed similar results with a rate significantly lower than that of the open group (OR, 0.49; 95% CI, 0.42–0.57) (Figure 2d). Specifically, minimally invasive approaches had lower rates of wound infections (OR, 0.39; 95% CI, 0.22–0.68), incisional hernias (OR, 0.20; 95% CI, 0.06–0.67), and ascites and cirrhotic decompensation events (OR, 0.50; 95% CI, 0.29–0.87) than the open group.

Forty studies reported data on postoperative mortality rates.^{10–20,22–37,39–43,45–49,53,54,56,57} There were no statistically significant differences between laparotomy and minimally invasive approaches for rates of both in-hospital mortality (OR, 1.01; 95% CI, 0.67–1.54) and postoperative mortality within 30 days of discharge (OR, 0.88; 95% CI, 0.41–1.88).

Cost Analysis

Eight studies included cost analyses and discussion on this outcome.^{10,28,38,40,44,46,55,58} Of these, one was excluded because it was out of scope.⁵⁸ Four studies reported cost differences between minimally invasive approaches and conventional approaches, with three comparing laparotomy with laparoscopy and one comparing a robotic approach with an open approach.^10,28,46,55 These studies showed a nonsignificant trend of higher total operative costs of $334.10 (95% CI, –$753.50–$1421.60) for minimally invasive approaches. Four studies reported total hospital cost differences, with all comparing laparotomy with laparoscopy.^28,40,46,55 These researchers found a trend of higher total hospital costs in patients undergoing the conventional open approach of $3223 (95% CI, –$474–$692). Of note, one additional article normalized cost values for both total operative costs and total hospital costs and was subsequently not included in the statistical analysis.³⁸

DISCUSSION

This meta-analysis of 3702 patients over a 14-year period yielded 49 pertinent studies showing minimally invasive approaches for hepatectomy to be as safe and efficacious as conventional laparotomy, with similar total operative times. Minimally invasive approaches afford shorter LOH, decreased EBL, and decreased postoperative morbidity. Specifically, these approaches resulted in fewer incisional hernias, wound infections, and ascites or cirrhotic decompensation events and retained oncologic integrity. All approaches to liver resection resulted in similar mortality rates. In terms of cost, minimally invasive approaches required nearly the same amount of money in the operating room as the conventional approach but saved money over the entire LOH.

Favorable operative outcomes, such as decreased EBL and lower rates of postoperative morbidity, lend credence to increased implementation of minimally invasive approaches. Bile leaks and massive hemorrhages are two important perioperative considerations in hepatic surgery owing to the unique anatomic structure of the liver, with minimally invasive approaches showing decreased intraoperative blood loss and equitable postoperative bile leak rates. The observed lower EBL is likely multifactorial, owing to both hepatic vein tamponade from pneumoperitoneum and improved dissection via field magnification. Furthermore, higher EBL and consequent blood transfusions are associated with increased postoperative morbidity, helping explain the lower rates of postoperative morbidity observed in this study.⁵⁹

Long-term mortality rates were reported for 22 of the included study samples.^{11,15,17,19,24,25,29,32,33,35,36,38,41,45,47,48,51,53,54,56,57} When immediate postoperative deaths were excluded, nonsignificant differences were found between laparoscopic hepatectomy and open hepatectomy for overall survival and for disease-free survival by all research groups except one. Kandil et al²⁴ found no difference in overall survival (P = .818) but found a significant difference in disease-free survival, with 100% 3-year survival in laparoscopic hepatectomy patients versus 71.4% survival in open hepatectomy patients (P = .03). Of note, the operative indication for this research group was neuroendocrine metastasis, whereas the indications for the remaining groups were primarily hepatocellular carcinoma or colorectal cancer metastases (Table 2). Perhaps a survival advantage exists in this population of patients; however, further studies are needed to establish the potential validity of this relationship.

A focus of debate regarding implementation of minimally invasive surgery centers on cost. In comparing total operative costs and total hospital costs among groups, studies found that although operative costs were higher for laparoscopic groups, their hospitalization costs were lower because of shorter LOH, which is intimately tied to postoperative morbidity, as well as decreased intensive care unit admission rates.^38,40,60 Only 1 article assessed comparative costs between robotic and conventional open approaches, finding increased operating room costs with the robotic approach.¹⁰ However, without discussion of total hospital costs, no conclusions can be drawn from that study regarding the potential financial tradeoff gained by implementing robotic intervention. Further studies including the economic impact of minimally invasive surgery are needed to advance this discussion.

Minimally invasive approaches to surgery afford the surgeon increased visibility and the patient decreased LOH, improved cosmesis, and decreased postoperative pain. Colorectal metastases are a leading indication for hepatectomy, for which a majority of patients need repeat hepatectomy. Minimally invasive approaches not only better facilitate reoperations in this patient population but also allow for simultaneous operations in colorectal cancer patients with synchronous hepatic metastases.^61–64

Although this study is comprehensive and is the most current evaluation of approaches to liver resection, there are several limitations and shortcomings to our study. First, the included studies are nonrandomized, retrospective studies, making them of moderate quality with increased selection bias. Also contributing to selection bias was patient selection by the surgeon, wherein healthier patients more fit for surgery were more likely to undergo minimally invasive options, leading to more favorable postoperative outcomes. Furthermore, patients selected for laparoscopic surgery may have had more easily resectable tumors, possibly contributing to their relative increase in negative margins. Intimately linked to minimally invasive surgical outcomes is both the surgeon's experience with the procedure and the volume of cases to which each care center is accustomed, neither of which was included in these studies, thereby prohibiting subanalysis. These studies exhibited moderate heterogeneity, with varying surgical techniques and differing outcome measures. Specifically, significant heterogeneity in reporting of resection outcomes, positive and negative versus R0–R1, prevents subanalysis of this outcome.

Although 873 citations were initially identified, an overwhelming majority of these were out of scope, focusing on tangential topics relating to liver donations, radiofrequency ablation, and tumor staging. Moreover, although these articles may have marginally touched on some of our primary outcomes, they neglected to contain data pertinent to this study. Furthermore, patient overlap by research groups led to the exclusion of 5 articles from analysis, totaling 488 patient experiences that are not represented.

The only statistically significant difference noted between minimally invasive approaches was a roughly 10% lower conversion rate to open surgery in the laparoscopic group as compared with the robotic group. With only 3 comparative studies including a robotic group, the ability to accurately ascertain any relationship to the robotic group is limited by its underpowering and the subsequent inability to perform subgroup analysis. Further comparative studies that include robotic approaches are needed. At present, the limited volume is likely because of the financial investment and operative training required to implement robots into common surgical practice.

CONCLUSION

To our knowledge, this review represents the largest, most current analysis of outcomes related to minimally invasive approaches to hepatectomy, with minimally invasive approaches showing improved postoperative morbidity, retained oncologic integrity, and potentially decreased economic burden to the health care system. Furthermore, future research comparing the robotic approach with the laparoscopic approach, as well as assessing the cost associated with each approach, is warranted.