Laparoscopic versus open hepatic resections for benign and malignant neoplasms—a meta-analysis

Laparoscopic versus open hepatic resections for benign and malignant neoplasms—a meta-analysis Constantinos Simillis, BSc, MBBS,a Vasilis A. Constantinides, BSc, MBBS,a Paris P. Tekkis, MD, FRCS,a Ara Darzi, MD, FRCS, KBE,a Richard Lovegrove, MRCS,a Long Jiao, MD, FRCS,b and Anthony Antoniou, FRCS,a London, United Kingdom

Background. Laparoscopic surgery for hepatic neoplasms aims to provide curative resection while minimizing complications. The present study compared laparoscopic versus open surgery for patients with hepatic neoplasms with regard to short-term outcomes. Methods. Comparative studies published between 1998 and 2005 were included. Evaluated endpoints were operative, functional, and adverse events. A random-effects model was used and sensitivity analysis performed to account for bias in patient selection. Results. Eight nonrandomized studies were included, reporting on 409 resections of hepatic neoplasms, of which 165 (40.3%) were laparoscopic and 244 (59.7%) were open. Operative blood loss (weighted mean difference ⫽ ⫺123 mL; confidence interval ⫽ ⫺179, ⫺67 mL) and duration of hospital stay (weighted mean difference ⫽ ⫺2.6 days; confidence interval ⫽ ⫺3.8, ⫺1.4 days) were significantly reduced after laparoscopic surgery. These findings remained consistent when considering studies matched for the presence of malignancy and segment resection. There was no difference in postoperative adverse events and extent of oncologic clearance. Conclusions. Laparoscopic resection results in reduced operative blood loss and earlier recovery with oncologic clearance comparable with open surgery. When performed by experienced surgeons in selected patients it may be a safe and feasible option. Because of the potential of significant bias arising from the included studies, further randomized controlled trials should be undertaken to confirm this bias and to assess long-term survival rates. (Surgery 2007;141:203-11.) From the Imperial College London, Department of Surgical Oncology and Technology, St Mary’s Hospital,a and the Imperial College London, Department of General and Hepatopancreatobiliary Surgery, Hammersmith Hospitalb

Laparoscopic liver resections have been looked upon with skepticism because of concerns regarding parenchymal transection, bleeding control, and bile leakage.1-3 There are further risks of incomplete resections and air embolism.2,3 Retraction and liver mobilization can be difficult, and there is potential danger of injuring major adjacent structures.4 Advances in laparoscopic instruments and increasing operator experience have led surgeons to explore the role of laparoscopy in liver resections. Improve-

Accepted for publication June 12, 2006. Reprint requests: Anthony Antoniou, FRCS, Consultant Colorectal Surgeon, Imperial College London, Department of Surgical Oncology and Technology, St Mary’s Hospital, 10th Flr, QEQM Wing, Praed St, London W2 1NY, UK. E-mail: ant.a@ virgin.net. 0039-6060/$ - see front matter © 2007 Mosby, Inc. All rights reserved. doi:10.1016/j.surg.2006.06.035

ments in laparoscopic instruments include hemostatic devices such as the ultrasonic dissector (harmonic scalpel),5 microwave coagulators, argon beam coagulators, together with laparoscopic ultrasound,6 endoscopic linear staplers, and development of hand-assisted laparoscopic techniques.4 Initial laparoscopic liver procedures included nonanatomic wedge biopsies7 and fenestration of hepatic cysts8, followed by nonanatomic and anatomic liver resections of benign or malignant neoplasms.9-13 Previous studies have demonstrated the feasibility and safety of laparoscopic liver resections, but all included small number of patients.10,12-14 In addition, the precise role of laparoscopy in resection of liver neoplasms (primary and metastatic) remained controversial, with only a few studies comparing short-term outcomes of laparoscopic versus conventional resection of liver neoplasms.2,3,15-20 The aim of the present study was to evaluate perioperative parameters and adverse outcomes in patients undergoing laparoscopic versus open hepSURGERY 203

204 Simillis et al

atectomy. Meta-analytical techniques were used to assess each technique and explain differences between the study methodologies and selection criteria. The limitations of this meta-analysis are also discussed. METHODS Study selection. A Medline, Embase, Ovid, and Cochrane database search was performed on all comparative studies of laparoscopic versus open approaches for resection of benign and malignant neoplasms in the liver. The following Mesh search headings were used: “laparoscopy,” “hepatectomy,” “comparative study,” and “treatment outcome.” The above terms and their combinations were also searched as text words, as were “laparoscopic,” “liver surgery,” “minimally invasive surgery,” and “liver resection.” The “related articles” function was used to broaden the search, and all abstracts, studies, and citations scanned were reviewed. References of the articles acquired were also searched by hand. No language restrictions were made. The latest date for this search was November 10, 2005. Data extraction. Two reviewers (C.S. and V.C.) independently extracted the following parameters from each study: first author, year of publication, study population characteristics, study design, inclusion and exclusion criteria, matching criteria, number of subjects operated on with each technique, male to female ratio, and conversion rates. There was 100% agreement between the 2 reviewers. Inclusion criteria. To be included in the analysis, studies had to (1) compare laparoscopic and open approaches in patients undergoing hepatectomy for benign and malignant neoplasms; (2) report on at least one of the outcome measures mentioned below; (3) clearly document the operative technique as “laparoscopic” or “open”; (4) clearly report the indications for surgery for each of the laparoscopic and open groups; (5) when two studies were reported by the same institution and/or authors, either the one of higher quality or the most recent publication was included in the analysis. Exclusion criteria. Studies were excluded from the analysis if (1) the outcomes of interest were not clearly reported; (2) it was impossible to extract or calculate the appropriate data from the published results; (3) there was considerable overlap between authors, centers, or patient cohorts evaluated in the published literature; (4) resections were carried out only for the purpose of cyst excision, biopsy, or nonanatomic resection. Outcomes of interest and definitions. The following outcomes were used to compare patients

Surgery February 2007

undergoing laparoscopic hepatic resection (LHR) with those undergoing open hepatic resection (OHR): ●

● ●

●

Operative parameters included operative time, operative blood loss, number of patients requiring blood transfusion, use of portal triad clamping, duration of portal triad clamping, and oncologic clearance in terms of pathologic resection margins. Postoperative parameter comprised time to first oral intake and duration of postoperative hospital stay. Early postoperative adverse events included postoperative complications divided into liver resection–related (cirrhotic decompensation/ascites, hepatic hemorrhage, biliary leakage) and general complications (chest infection, incisional hernia, bowel perforation, urinary infection). Long-term outcomes included survival at maximal follow-up and recurrence-free survival at maximal follow-up.

Statistical analysis. The meta-analysis was performed in line with recommendations from the Cochrane Collaboration and the Quality of Reporting of Meta-analyses (QUORUM) guidelines.21,22 Statistical analysis of dichotomous variables was carried out by using odds ratio (OR) as the summary statistic, while continuous variables were analyzed using the weighted mean difference (WMD),23 and both were reported with 95% confidence intervals (CIs). Odds ratios represent the odds of an adverse event occurring in the LHR group compared with the OHR group, while WMDs summarize the differences between the 2 groups with respect to continuous variables, accounting for sample size. For studies that presented continuous data as means and range values, the standard deviations (SDs) were calculated by using statistical algorithms and checked by using “bootstrap” resampling techniques. Thus all continuous data were standardized for analysis. An OR of less than 1 favored the LHR group, and the point estimate of the OR was considered to be statistically significant at the P ⬍ .05 level if the 95% CI did not include the value “1.” The Mantel-Haenszel method was used to combine the OR for the outcomes of interest by using a “random effect” meta-analytical technique. In a random effect model it is assumed that there is variation between studies and the calculated odds ratio, and thus has a more conservative value.23,24 The random effect model is preferable when metaanalytical techniques are used in surgical research for a given surgical technique in which each centre has its own patient selection criteria and these patients have different risk profiles. The Yate correction was used for those studies that contained a

Surgery Volume 141, Number 2

zero in 1 cell for the number of events of interest in 1 of the 2 groups.25,26 These “zero cells” created problems with the computation of ratio measure and its standard error of the treatment effect. These problems were resolved by adding the value “0.5” in each cell of the 2 ⫻ 2 table for the study in question; furthermore, if there were no events for both laparoscopic and open groups, the study was discarded from the meta-analysis of that outcome. The quality of the studies was assessed by using the Newcastle-Ottawa Scale (NOS) with some modifications to match the needs of this study.27 The quality of the studies was evaluated by examining 3 factors: patient selection, comparability of the study groups, and assessment of outcome. Studies achieving 6 or more stars were considered to be of higher quality. Heterogeneity was assessed by 2 methods. First, graphical exploration with funnel plots was used to evaluate publication bias.24,28 Second, sensitivity analysis was undertaken by using the following subgroups: studies of higher quality,3,16,18-20 those published in or after 2003,2,3,18-20 those reporting on more than 20 resections in the laparoscopic group,3,17,19 studies matched for the presence of malignancy,15,16,19,20 and studies matched for segment resection.3,15,17,20 Analysis was conducted by using the statistical software Review Manager, version 4.2 (The Cochrane Collaboration, Software Update, Oxford, United Kingdom). RESULTS Eligible studies. The literature search identified 13 studies1-4,14-20,29,30 that compared the results of LHR to OHR for benign and malignant liver neoplasms. Of these, 3 studies1,4,14 were noncomparative, and 1 study29 was a leading article with no reported outcomes; all 4 studies were excluded. Another study30 was excluded because the LHR group comprised only 2 cases of nonresectional surgery. Eight studies2,3,15-20 published between 1998 and 2005 matched the selection criteria and were therefore included. Analysis was performed on 403 patients who underwent 409 resections of hepatic neoplasms, 165 (40.3%) of which were laparoscopic and 244 (59.7%) were open. The study characteristics and patient demography are shown in Table I. Conversion to open surgery was reported in 6 (3.7%) cases among 4 studies,15,18-20 and they were analyzed as intention to treat. In one study17 only benign neoplasms were resected. In another study16 only colorectal metastases were resected, while in 2 other studies18,19 only hepatocellular carcinomas were resected. In most studies different segments

Simillis et al 205

of the liver were resected, except for a study20 in which only Couinaud segments II and III were resected, and another study15 in which resections included Couinaud segments II or III in 32 of the 34 (94%) resections. Details of the neoplasms resected in the open and laparoscopic group of each study are shown in Table II, which is available in the electronic version online. Five studies included matched patients, 1 with prospective15 and 4 with retrospective3,17,18, 20 data collection for the open group. In these studies, patients in the 2 groups were matched for gender,3,15,17 age,3,15,17 American Society of Anesthesiologists (ASA) status,3 body mass index (BMI),17 the presence or absence of cirrhosis,3,20 severity of liver disease (Child-Pugh criteria, esophageal varices, and fibrosis histologic score),18 type of liver resection performed,3,17,18 neoplasm histology,15,20 size,3,15,17,18,20 and location.3,15,17,20 Results from the meta-analysis of the studies with regard to operative and postoperative parameters and adverse events for laparoscopic versus open group are summarized in Table III, which is available as an electronic version online. Figure 1, which is also available as an electronic version online, is a forest plot displaying the results of the analysis for operative blood loss, duration of hospital stay, and overall postoperative complications. Meta-analysis of operative parameters. Seven studies3,15-20 reported on operative blood loss, which was found to be significantly lower in the LHR group versus the OHR group by 123 mL (95% CI, ⫺179, ⫺67; P ⬍ .001), and this finding was not associated with significant heterogeneity between studies. No significant difference was shown between the groups regarding the need for blood transfusions, operative time, or incidence of portal triad clamping. Two studies18,20 reported on duration of portal triad clamping, which was found to be significantly longer in the laparoscopic group by 28 minutes (95% CI, 2, 55; P ⫽ .03). Four studies3,16,18,20 reported on the adequacy of oncologic clearance in terms of pathologic resection margins with no significant difference between the 2 groups. Meta-analysis of postoperative parameters. All studies reported on the duration of hospital stay, which was found to be significantly less in the laparoscopic group by 2.6 days (95% CI, ⫺3.8, ⫺1.4, P ⬍ .001), without significant heterogeneity between the studies. The period to first oral intake was also significantly less in the laparoscopic group by 0.5 days (95% CI, ⫺1.0, 0.0, P ⫽ .05), but there was a marginally significant heterogeneity between the studies for this outcome (P ⫽ .04).

206 Simillis et al

Table I. Study characteristics— demographics No. of patients Author

Year

Morino et al

2003

Lesurtel et al

Design

Matching*

Inclusion criteria†

Exclusion criteria‡

Conversions (n [%])

Mean age (median)

Female (n [%])

0

L ⫽ 56 O ⫽ 58 L ⫽ 55 O ⫽ 47 L ⫽ (68) O ⫽ (59) n/c

L ⫽ 16 (53) O ⫽ 19 (63) L ⫽ 11 (61) O ⫽ 15 (75) L ⫽ 4 (31) O ⫽ 4 (29) L ⫽ 21 (100) O ⫽ 21 (100) L ⫽ 10 (59) O ⫽ n/c L ⫽ 12 (40) O ⫽ 18 (64) L ⫽ 3 (23) O ⫽ 4 (29) n/c

L

O

RM

30

30

1-3, 5-9

1-3

3

2003

P(L), RM(O)

18

20

1-8, 13

1-4

3, 4

2 (11.1)

Mala et al

2002

R

13

14

1-6, 9, 12, 13

1

2, 5

0

Farges et al

2002

P(L), RM(O)

21

21

1, 2, 5-7, 14

2, 5

1, 5, 6

0

Buell et al

2004

P(L), R(O)

17

100

5

1, 2, 4

—

0

Kaneko et al

2005

P(L), R(O)

30

28

1, 2, 15-20

1, 4

2, 7

1 (3.3)

Laurent et al

2003

P(L), RM(O)

13

14

1, 4, 6

2, 3, 7

2 (15.4)

Rau et al

1998

PM

17

17

1, 2, 4-8, 10, 13, 15-17, 19, 21-24 1, 2, 4-6, 13, 25

1, 2

3, 5

1 (5.9)

L ⫽ 55.4 O ⫽ n/c L ⫽ 59 O ⫽ 61 L ⫽ 62.6 O ⫽ 65.9 L ⫽ 48 O ⫽ 46.8

Study quality (star rating) (max: 10) ******** ****** ****** ***** ***** ****** ****** *****

L, Laparoscopic; max, maximum; O, open; n/c, not commented; P, prospective; PM, prospective matched; R, retrospective; RM, retrospective matched. *1 ⫽ age; 2 ⫽ gender; 3 ⫽ American Society of Anesthesiologists (ASA) classification; 4 ⫽ malignancy; 5 ⫽ mean size of lesion; 6 ⫽ location of neoplasm; 7 ⫽ resection type; 8 ⫽ cirrhosis; 9 ⫽ liver metastases; 10 ⫽ primary malignancy; 11 ⫽ associated procedures during liver resection; 12 ⫽ previous operations; 13 ⫽ neoplasm histology; 14 ⫽ BMI; 15 ⫽ Child-Pugh grading. †1 ⫽ malignancy; 2 ⫽ benign disease; 3 ⫽ left lobectomies (segments II; III and IV); 4 ⫽ primary hepatic cancer; 5 ⫽ only noncirrhotic patients; 6 ⫽ only patients with chronic liver disease. ‡1 ⫽ malignancy; 2 ⫽ benign disease; 3 ⫽ lesions in segments VII; VIII; 4 ⫽ right lobectomies; 5 ⫽ primary hepatic cancer; 6 ⫽ cirrhosis; 7 ⫽ metastatic cancer.

Surgery February 2007

Surgery Volume 141, Number 2

Meta-analysis of adverse events. No significant differences were found between the 2 surgical approaches regarding postoperative adverse events. These included overall complications (LHR, 12%; OHR, 17%; P ⫽ .24), complications related directly to the liver resection (LHR, 5%; OHR, 13%; P ⫽ .09), and general complications (LHR, 8%; OHR, 4%; P ⫽ .38). Specific complications analyzed included chest infections (LHR, 8%; OHR, 3%; P ⫽ .31), cirrhotic decompensation/ascites (LHR, 4%; OHR, 10%; P ⫽ .22), hepatic hemorrhage (LHR, 2%; OHR, 4%; P ⫽ .70), and biliary leakage (LHR, 4%; OHR, 6%; P ⫽ .55). Postoperative mortality due to hepatic failure was reported in only 1 patient in the laparoscopic group,2 and 2 patients in the open group.18 Sensitivity analysis. Results of the analysis are shown in Table IV, which is also available online. High-quality studies (ⱖ6 stars): Analysis of the high-quality studies3,16,18-20 showed the duration of stay to be significantly shorter (WMD, ⫺3.2 days; 95% CI, ⫺4.9, ⫺1.6; P ⬍ .001), the operative blood loss to be significantly less (WMD, ⫺149 mL, 95% CI, ⫺217, ⫺80, P ⬍ .001), and a significant decrease in the complications related to liver resection (OR, 0.2; 95% CI, 0.1, 0.8; P ⫽ .02) in the LHR group. Studies published in or after 2003: More recent reports2,3,18-20 revealed a significantly reduced operative blood loss (WMD, ⫺151 mL; 95% CI, ⫺220, ⫺82; P ⬍ .001), a significantly shorter duration of stay (WMD, –3.0 days, 95% CI, ⫺4.5, ⫺1.4; P ⬍ .001), and a significant decrease in the complications related to liver resection (OR, 0.2; 95% CI, 0.1, 0.9; P ⫽ .03) in the LHR group. Studies reporting on 20 or more resections in the laparoscopic group: Analysis of these studies3,17,19 revealed a significantly reduced operative blood loss (WMD, ⫺117 mL, 95% CI, ⫺187, ⫺47; P ⫽ .001) and a shorter duration of hospital stay (WMD, ⫺2.8 days; 95% CI, ⫺4.7, ⫺0.9; P ⫽ .004). Studies matched for the presence of malignancy: Similar to previous sensitivity analysis, evaluation of studies matched for the histology of the neoplasm resected15,16,19,20 revealed a significantly reduced operative blood loss (WMD, ⫺157 mL, 95% CI, ⫺235, ⫺79, P ⬍ .001) and a significantly shorter duration of hospital stay (WMD, ⫺3.9 days; 95% CI, ⫺6.1, ⫺1.8, P ⬍ .001) in the laparoscopic group. Studies matched for segment resection: The LHR group in studies that were matched for segment resections3,15,17,20 exhibited significant reductions in operative blood loss (WMD, ⫺115 mL; 95% CI, ⫺191, ⫺38; P ⫽ .003) and duration of hospital stay (WMD, ⫺1.5 days; 95% CI, ⫺2.2, ⫺0.9; P ⬍ .001),

Simillis et al 207

but there was a significant increase in operative time (WMD, 45 minutes; 95% CI, 22, 68; P ⬍ .001). Other outcomes of interest: Two studies16,17 compared the postoperative need for analgesia between the laparoscopic and open group. Mala et al16 found a significant decrease in the postoperative need of analgesia in the laparoscopic group (1 vs 5 days). Farges et al17 assessed the analgesia requirement between postoperative days 1 and 3; postoperative pain was relieved by morphine and acetaminophen. Cumulative morphine requirement (but not acetaminophen) was found to be significantly lower after laparoscopic surgery because of a decreased requirement on postoperative day 1. Three studies16,18,19 included long-term follow-up of the patients. In one study,16 2 patients in the laparoscopic group died 11 and 17 months postoperatively, and 3e patients in the open group died 11, 18, and 40 months postoperatively. Kaneko et al19 observed no recurrences related to laparoscopy, such as peritoneal dissemination or port-site recurrences. Also, there were no significant differences between procedures concerning the 5-year survival rate (laparoscopic, 61%; open, 62%) and 5-year survival rate without recurrences (laparoscopic, 31%; open, 29%). Finally, Laurent et al18 revealed a similar 3-year recurrence-free survival between the 2 groups (laparoscopic, 44%; open, 46%) and a significantly higher 3-year survival in the laparoscopic group (89% vs 55%; P ⫽ .04) which was suppressed when the 2 postoperative deaths in the open group were excluded. No patient, in either group, developed recurrence at the site of resection, and no port-site recurrences were observed in the laparoscopic group. DISCUSSION The present study suggested that LHR may be as safe as OHR for selected patients, with comparable results. Furthermore, LHR was found to be associated with lower operative blood loss and reduced hospital stay, with potentially significant implications in terms of health care costs. This finding was consistent throughout all sensitivity analyses. However, careful interpretation of the results is required in the light of significant patient selection bias arising from the retrospective nature of the included studies, and potential confounding factors that are inherent to the study design and limit extrapolation of the results to the clinical setting. By decreasing trauma, laparoscopic surgery has been shown to result in reduced postoperative pain, shorter hospital stay, faster return to

208 Simillis et al

normal activity, and cosmetic benefits.13,16,17,31,32 These benefits in turn may improve tolerance and acceptance for reoperations, such as repeat hepatectomy for recurrent cancer or subsequent liver transplantation.16 Other advantages in cancer patients may include greater preservation of the immune function, with some authors advocating possible enhancement in the antineoplasm response1,33 and earlier access to adjuvant treatment by earlier recovery.10 Moreover, laparoscopic surgery may provide better visualization of deep vascular structures and possibly more precise and accurate surgery.1 At present LHR is mainly done in a minority of cases at major institutions with a large liver resection caseload because laparoscopic hepatectomy is a complex procedure requiring experience and different skills than open surgery because of a 2-dimensional representation of the operative site, limited tactile feedback, and the need for surgeons to learn different hand-eye coordination skills. The training of surgeons, together with the cost of the equipment and the possible increased demand for operations, may act to increase costs, which may balance out any possible cost-effectiveness that could be offered by laparoscopic hepatectomy. In our study the operating time was comparable between LHR and OHR. This finding persisted in all subgroups, except for studies matched for segment resection, which showed the operating time to be significantly longer in the laparoscopic group. The increased operative time observed in some studies15,17,20 with the laparoscopic group may be, at least in part, due to the effect of the learning curve, but this observation could not be evaluated in the present study. LHR, when performed by a specialized surgical team, appears to be safe, with a complication rate similar to the open procedure. Even in cirrhotic patients, in whom liver resections carried a high risk of complications, such as the development of ascites, jaundice, and encephalopathy,34,35 complication rates were comparable between the 2 groups.3,18,20 Studies have suggested that the laparoscopic approach might improve the postoperative course of cirrhotic patients for the following reasons: preservation of the abdominal wall and round ligament avoids interruption of collateral circulation, less mobilization and manipulation of the liver reduces liver trauma, nonexposure of abdominal viscera restricts fluid requirements and decreases electrolytic and protein losses, and decreased intraoperative blood loss reduces the risk of postoperative death in cirrhotic patients.11,18,20,35,36 Nonetheless, at present,

Surgery February 2007

open hepatectomy has become very safe with a mortality rate decreased to 2% and a morbidity rate to less than 30%.37,38 One of the main concerns during hepatectomy is minimizing blood loss39 and avoidance of blood transfusions,40 which may be achieved by hypotensive anesthesia and vascular clamping, despite the drawback of ischemic (and/or reperfusion) injury to the liver parenchyma.41,42 Intraoperative blood loss was consistently lower in the LHR group than in the OHR group, but this finding was not clinically manifested in the need for blood transfusions, which was comparable between the groups. The Pringle maneuver was necessary for longer duration in the laparoscopic group, a fact that may partially explain the improvement in operative blood loss, together with the use of the ultrasonic scalpel, and the hemostatic effect of the pneumoperitoneum.3,20 In considering patients for LHR, the size, type (malignant or benign), location of the neoplasm, as well as the physiologic status of the patient and operator experience must be taken into account. One of the main indications of laparoscopy would be for patients requiring wedge resections of superficially or peripherally located neoplasms, whereas, for patients requiring segmentectomies or bisegmentectomies, the decision for laparoscopy is more difficult. In patients with neoplasms located in the left liver lobe and the anterior and inferior liver segments (IV anterior, V, and VI), LHR has been shown to be feasible and safe.9,20 On the other hand, neoplasms located in the right lobe and the posterior and superior liver segments (VII, VIII, IV posterior, and I) are technically more demanding and should be approached with caution.19,20 Moreover, the size of the neoplasm is important in selecting patients for laparoscopic hepatectomy, in which an acceptable diameter for nodular and pedunculated neoplasms would be 40 mm and 60 mm, respectively.3,19,43 When resecting a malignant neoplasm, it is imperative to achieve a disease-free resection margin, 44 which is difficult during LHR because of the loss of tactile sensation. The results of this study suggested that the extent of oncologic resection was comparable between LHR and OHR. A 1-cm free surgical margin should be obtained45,46 by using laparoscopic ultrasonography.14,47 There are also concerns regarding portsite metastases, wound recurrence, peritoneal seeding, and neoplasm growth promotion by carbon dioxide gas insufflation during laparoscopic procedures.14,48-51 There is no definitive evi-

Simillis et al 209

Surgery Volume 141, Number 2

dence in the literature that the use of the laparoscopic technique increases the risk of neoplasm dissemination,48,49,52 and none of the patients followed up in this study developed any port-site or cutaneous metastasis. Furthermore, LHR carries the potential risk of gas embolism caused by pneumoperitoneum during hepatic vein division,53,54 but no cases were reported in this meta-analysis. This potential complication seems to be rare, with a review by Biertho et al55 reporting on 2 cases of possible gas embolisms (1.1%) after 186 laparoscopic hepatectomies. This meta-analysis of nonrandomized studies may have several limitations that must be taken into account when considering its results. The role of metaanalytic techniques in this setting has been a source of extensive debate amongst epidemiologists. Critics argue that the effect of meta-analysis is to “reinforce the inherent systematic biases of the studies, produce spurious statistical stability and discourage further research.”56,57 Proponents of meta-analysis argue that statistical quantification and pooling of results from many studies provide an “excellent tool for identifying reasons for variability and inconsistency” and that the finding of heterogeneity “sets the stage for further research” on a given topic.58 Furthermore, distinction is made between “synthetic” meta-analysis, in which a single summary is reported while heterogeneity is ignored, and “comparative” meta-analysis, in which heterogeneity is taken into account and attempts are made to explain it.59 The latter serves as an aid in “critical comparison between studies,” and there seems to be general agreement that the “comparative” approach has a place in medical literature and may complement qualitative reviews.56,60 Recently, Deeks et al61 have evaluated nonrandomized intervention studies by using resampling techniques, and have found that “results of randomized and nonrandomized studies sometimes, but not always, differ and that both similarities and differences may often be explicable by other confounding factors. Part of the conclusion was that nonrandomized studies should only be undertaken when randomization is “unfeasible or unethical.” In the light of the above discussion, we believe that meta-analysis of nonrandomized studies is useful in the absence of randomized, controlled trials as well as to guide further researchers toward properly informed randomization in future studies. CONCLUSION The present study has shown that laparoscopic hepatectomy was associated with reduced operative blood loss, earlier postoperative recovery, and a resected specimen that was oncologically comparable with open surgery. When performed

by suitably specialized surgeons in selected patients, it appears to be a safe and feasible alternative to open surgery. However, currently there are only a small number of comparative, nonrandomized studies published that limit extrapolation of the results to the clinical setting. Results from randomized trials should be awaited with particular interest on the above-mentioned outcomes, but also extended follow-up is required to assess long-term survival rates before any definitive conclusions can be made. We acknowledge the advice of Mr Thanos Athanasiou regarding the statistical methodology and layout of the manuscript. REFERENCES 1. Vibert E, Perniceni T, Levard H, Denet C, Shahri NK, Gayet B. Laparoscopic liver resection. Br J Surg 2006;93:67-72. 2. Buell JF, Thomas MJ, Doty TC, et al. An initial experience and evolution of laparoscopic hepatic resectional surgery. Surgery 2004;136(4):804-11. 3. Morino M, Morra I, Rosso E, Miglietta C, Garrone C. Laparoscopic vs open hepatic resection: a comparative study. Surg Endosc 2003;17(12):1914-8. 4. Fong Y, Jarnagin W, Conlon KC, DeMatteo R, Dougherty E, Blumgart LH. Hand-assisted laparoscopic liver resection: lessons from an initial experience. Arch Surg 2000;135(7): 854-9. 5. Rothenberg SS. Laparoscopic splenectomy using the harmonic scalpel. J Laparoendosc Surg 1996;6(Suppl 1):S61-3. 6. Cuesta MA, Meijer S, Borgstein PJ, Sibinga Mulder L, Sikkenk AC. Laparoscopic ultrasonography for hepatobiliary and pancreatic malignancy. Br J Surg 1993;80(12):1571-4. 7. Falcone RE, Wanamaker SR, Barnes F, Baxter CG, Santanello SA. Laparoscopic vs. open wedge biopsy of the liver. J Laparoendosc Surg 1993;3(4):325-9. 8. Morino M, De Giuli M, Festa V, Garrone C. Laparoscopic management of symptomatic nonparasitic cysts of the liver. Indications and results. Ann Surg 1994;219(2):157-64. 9. Mouiel J, Katkhouda N, Gugenheim J, Fabiani P. Possibilities of laparoscopic liver resection. J Hepatobiliary Pancreat Surg 2000;7(1):1-8. 10. Cherqui D, Husson E, Hammoud R, et al. Laparoscopic liver resections: a feasibility study in 30 patients. Ann Surg 2000; 232(6):753-62. 11. Abdel-Atty MY, Farges O, Jagot P, Belghiti J. Laparoscopy extends the indications for liver resection in patients with cirrhosis. Br J Surg 1999;86(11):1397-400. 12. Kaneko H, Takagi S, Shiba T. Laparoscopic partial hepatectomy and left lateral segmentectomy: technique and results of a clinical series. Surgery 1996;120(3):468-75. 13. Huscher CG, Lirici MM, Chiodini S. Laparoscopic liver resections. Semin Laparosc Surg 1998;5(3):204-10. 14. Gigot JF, Glineur D, Santiago Azagra J, et al. Laparoscopic liver resection for malignant liver tumors: preliminary results of a multicenter European study. Ann Surg 2002; 236(1):90-7. 15. Rau HG, Buttler E, Meyer G, Schardey HM, Schildberg FW. Laparoscopic liver resection compared with conventional partial hepatectomy—a prospective analysis. Hepatogastroenterology 1998;45(24):2333-8. 16. Mala T, Edwin B, Gladhaug I, et al. A comparative study of

210 Simillis et al

17.

18.

19.

20.

21.

22.

23. 24.

25.

26.

27.

28. 29. 30.

31. 32.

33.

34.

35.

36.

the short-term outcome following open and laparoscopic liver resection of colorectal metastases. Surg Endosc 2002; 16(7):1059-63. Farges O, Jagot P, Kirstetter P, Marty J, Belghiti J. Prospective assessment of the safety and benefit of laparoscopic liver resections. J Hepatobiliary Pancreat Surg 2002;9(2):242-8. Laurent A, Cherqui D, Lesurtel M, Brunetti F, Tayar C, Fagniez PL. Laparoscopic liver resection for subcapsular hepatocellular carcinoma complicating chronic liver disease. Arch Surg 2003;138(7):763-9; discussion 769. Kaneko H, Takagi S, Otsuka Y, et al. Laparoscopic liver resection of hepatocellular carcinoma. Am J Surg 2005; 189(2):190-4. Lesurtel M, Cherqui D, Laurent A, Tayar C, Fagniez PL. Laparoscopic versus open left lateral hepatic lobectomy: a case-control study. J Am Coll Surg 2003;196(2):236-42. Clarke M, Horton R. Bringing it all together: LancetCochrane collaborate on systematic reviews. Lancet 2001; 357(9270):1728. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 2000;283(15):2008-12. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7(3):177-88. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997;315(7109):629-34. Yusuf S, Peto R, Lewis J, Collins R, Sleight P. Beta blockade during and after myocardial infarction: an overview of the randomized trials. Prog Cardiovasc Dis 1985;27(5):335-71. Mantel N, Haenszel W. Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst 1959;22(4):719-48. Athanasiou T, Al-Ruzzeh S, Kumar P, et al. Off-pump myocardial revascularization is associated with less incidence of stroke in elderly patients. Ann Thorac Surg 2004;77(2): 745-53. Egger M, Smith GD. Misleading meta-analysis. BMJ 1995; 311(7007):753-4. Cherqui D. Laparoscopic liver resection. Br J Surg 2003; 90(6):644-6. Trupka A, Hallfeldt K, Kalteis T, Schmidbauer S, Schweiberer L. [Open and laparoscopic liver resection with a new ultrasound scalpel]. Chirurg 1998;69(12):1352-6. Kohler L. Endoscopic surgery: what has passed the test? World J Surg 1999;23(8):816-24. Dunker MS, Stiggelbout AM, van Hogezand RA, Ringers J, Griffioen G, Bemelman WA. Cosmesis and body image after laparoscopic-assisted and open ileocolic resection for Crohn’s disease. Surg Endosc 1998;12(11):1334-40. Burpee SE, Kurian M, Murakame Y, Benevides S, Gagner M. The metabolic and immune response to laparoscopic versus open liver resection. Surg Endosc 2002;16(6):899-904. Bruix J, Castells A, Bosch J, et al. Surgical resection of hepatocellular carcinoma in cirrhotic patients: prognostic value of preoperative portal pressure. Gastroenterology 1996;111(4):1018-22. Takenaka K, Kanematsu T, Fukuzawa K, Sugimachi K. Can hepatic failure after surgery for hepatocellular carcinoma in cirrhotic patients be prevented? World J Surg 1990;14(1):123-7. Wu CC, Hwang CR, Liu TJ, P’Eng FK. Effects and limitations of prolonged intermittent ischaemia for hepatic resection of the cirrhotic liver. Br J Surg 1996;83(1):121-4.

Surgery February 2007

37. Taniguchi H, Takahashi T. Analysis of 210 elective hepatic resections. Hepatogastroenterology 1997;44(18):1624-31. 38. Buell JF, Rosen S, Yoshida A, et al. Hepatic resection: effective treatment for primary and secondary tumors. Surgery 2000;128(4):686-93. 39. Nagorney DM, van Heerden JA, Ilstrup DM, Adson MA. Primary hepatic malignancy: surgical management and determinants of survival. Surgery 1989;106(4):740-8; discussion 748-9. 40. Makuuchi M, Takayama T, Gunven P, Kosuge T, Yamazaki S, Hasegawa H. Restrictive versus liberal blood transfusion policy for hepatectomies in cirrhotic patients. World J Surg 1989;13(5):644-8. 41. Huguet C, Gavelli A, Chieco PA, et al. Liver ischemia for hepatic resection: where is the limit? Surgery 1992;111(3):251-9. 42. Belghiti J, Noun R, Malafosse R, et al. Continuous versus intermittent portal triad clamping for liver resection: a controlled study. Ann Surg 1999;229(3):369-75. 43. Takagi S, Kaneko H, Ishii T, et al. Laparoscopic hepatectomy for extrahepatic growing tumor. Surgical strategy based on extrahepatic growing index. Surg Endosc 2002;16(11):1573-8. 44. Nordlinger B, Guiguet M, Vaillant JC, et al. Surgical resection of colorectal carcinoma metastases to the liver. A prognostic scoring system to improve case selection, based on 1568 patients. Association Francaise de Chirurgie. Cancer 1996;77(7):1254-62. 45. Masutani S, Sasaki Y, Imaoka S, et al. The prognostic significance of surgical margin in liver resection of patients with hepatocellular carcinoma. Arch Surg 1994;129(10):1025-30. 46. Shirabe K, Takenaka K, Gion T, et al. Analysis of prognostic risk factors in hepatic resection for metastatic colorectal carcinoma with special reference to the surgical margin. Br J Surg 1997;84(8):1077-80. 47. Rahusen FD, Cuesta MA, Borgstein PJ, et al. Selection of patients for resection of colorectal metastases to the liver using diagnostic laparoscopy and laparoscopic ultrasonography. Ann Surg 1999;230(1):31-7. 48. Stocchi L, Nelson H. Wound recurrences following laparoscopic-assisted colectomy for cancer. Arch Surg 2000;135 (8):948-58. 49. Allardyce RA. Is the port site really at risk? Biology, mechanisms and prevention: a critical view. Aust N Z J Surg 1999;69(7):479-85. 50. Gutt CN, Kim ZG, Hollander D, Bruttel T, Lorenz M. CO2 environment influences the growth of cultured human cancer cells dependent on insufflation pressure. Surg Endosc 2001;15(3):314-8. 51. Volz J, Koster S, Spacek Z, Paweletz N. The influence of pneumoperitoneum used in laparoscopic surgery on an intraabdominal tumor growth. Cancer 1999;86(5):770-4. 52. Takiguchi S, Matsuura N, Hamada Y, et al. Influence of CO2 pneumoperitoneum during laparoscopic surgery on cancer cell growth. Surg Endosc 2000;14(1):41-4. 53. Croce E, Azzola M, Russo R, Golia M, Angelini S, Olmi S. Laparoscopic liver tumour resection with the argon beam. Endosc Surg Allied Technol 1994;2(3-4):186-8. 54. Takagi S. Hepatic and portal vein blood flow during carbon dioxide pneumoperitoneum for laparoscopic hepatectomy. Surg Endosc 1998;12(5):427-31. 55. Biertho L, Waage A, Gagner M. [Laparoscopic hepatectomy]. Ann Chir 2002;127(3):164-70. 56. Shapiro S. Is there is or is there ain’t no baby? Dr. Shapiro replies to Drs. Petitti and Greenland. Am J Epidemiol 1994;140(9):788-91.

Surgery Volume 141, Number 2

57. Shapiro S. Meta-analysis/Shmeta-analysis. Am J Epidemiol 1994;140(9):771-8. 58. Petitti DB. Of babies and bathwater. Am J Epidemiol 1994;140(9):779-82. 59. Greenland S. Can meta-analysis be salvaged? Am J Epidemiol 1994;140(9):783-7.

Simillis et al 211

60. Longnecker MP. Re: “Point/counterpoint: meta-analysis of observational studies”. Am J Epidemiol 1995;142(7): 779-82. 61. Deeks JJ, Dinnes J, D’Amico R, et al. Evaluating non-randomised intervention studies. Health Technol Assess 2003; 7(27):iii-x, 1-173.

211.e1 Simillis et al

Table II. Study characteristics—resection details with the open and laparoscopic procedures No. of procedures (%) Author

L

O

Morino et al

30 (50)

30 (50)

Lesurtel et al Mala et al Farges et al

18 (47) 15 (52) 21 (50)

20 (53) 14 (48) 21 (50)

Buell et al

21 (17)

100 (83)

Kaneko et al Laurent et al Rau et al

30 (52) 13 (48) 17 (50)

28 (48) 14 (52) 17 (50)

Laparoscopic Pathology of lesions resected* B ⫽ 16, C ⫽ 10, D ⫽ 4, E ⫽ 3, F ⫽ 1, a ⫽ 5, b ⫽ 4, c ⫽ 1, d ⫽ 6 A ⫽ 6, B ⫽ 12 C ⫽ 13 B ⫽ 21, a ⫽ 2, b ⫽ 2, d ⫽ 17 C ⫽ 1, D ⫽ 4, E ⫽ 3, F ⫽ 1, a ⫽ 1, b ⫽ 2, c ⫽ 7, d ⫽ 2, e ⫽ 1 E ⫽ 30 E ⫽ 13 C ⫽ 2, b ⫽ 7, c ⫽ 3, d⫽5

Mean size (mm)

Open Procedures performed†

42

W ⫽ 5, S ⫽ 12, B ⫽ 13

46 26 49

B ⫽ 18 S ⫽ 6, B ⫽ 7, T ⫽ 2 W ⫽ 9, S ⫽ 4, B ⫽ 8, LLS ⫽ 6 S ⫽ 4, B ⫽ 5, T ⫽ 4, LLS ⫽ 7, RHL ⫽ 1

76

30 33.5 n/c

LLS ⫽ 10, PH ⫽ 20 W ⫽ 3, S ⫽ 7, B ⫽ 3 S ⫽ 9, B ⫽ 8

Pathology of lesions resected*

Mean size (mm)

Procedures performed†

B ⫽ 5, C ⫽ 13, D ⫽ 12, E ⫽ 9, G ⫽ 3, a ⫽ 1, b ⫽ 2, d ⫽ 2 A ⫽ 7, B ⫽ 13 C ⫽ 14 B ⫽ 21

41

W ⫽ 5, S ⫽ 12, B ⫽ 13

53 30 55

B ⫽ 100

87

B ⫽ 20 S ⫽ 9, B ⫽ 3, T ⫽ 2 W ⫽ 9, S ⫽ 4, B ⫽ 8, LLS ⫽ 7 n/c

E ⫽ 28

31 34.3 n/c

LLS ⫽ 8, PH ⫽ 20 W ⫽ 4, S ⫽ 7, B ⫽ 3 S ⫽ 6, B ⫽ 10, T ⫽ 1

C ⫽ 2, b ⫽ 7, c ⫽ 3, d⫽5

L, Laparoscopic; n/c, not commented; O, open. *A, Malignant; B, benign; C, malignant metastatic; D, malignant primary; E, hepatocellular carcinoma; F, lymphoma; G, gallbladder cancer; a, adenoma; b, hemangioma; c, hydatid or other cysts; d, focal nodular hyperplasia; e, granuloma. †W, Wedge resections; S, segmentectomies; B, bisegmentectomies; T, three segments resection; LLS, left lateral segmentectomy; RHL, right hepatic lobectomy; PH, partial hepatectomy; n/c ⫽ not commented.

Surgery February 2007

Simillis et al 211.e2

Surgery Volume 141, Number 2

Table III. Results of meta-analysis comparing laparoscopic versus open hepatectomy Outcome of interest Operative parameters Operative time (min) Operative blood loss (mL) No. of patients requiring blood transfusion Use of portal triad clamping Duration of portal triad clamping (min) Pathologic resection margins Invaded by neoplasm Clear by ⬍1 cm Clear by ⬎1 cm Postoperative adverse events Overall complications Liver resection related complications General complications Chest infection Cirrhotic decompensation/ascites Hepatic hemorrhage Biliary leakage Postoperative parameters Duration of hospital stay (d) Time to first oral intake (d)

No. of studies

No. of resections

OR/WMD

95% CI

P value*

HG P value*

6 7 5 4 2

228 288 201 154 65

31.75 ⫺123.24 0.63 0.58 28.46

⫺6.54, 70.04 ⫺179.35, ⫺67.13 0.16, 2.49 0.09, 3.72 2.08, 54.84

.10 ⬍.001 .51 .56 .03

⬍.001 .81 .24 .01 .003

4 3 4

108 79 108

0.47 1.27 1.04

0.06, 3.49 0.44, 3.69 0.43, 2.52

.46 .66 .93

.90 .77 .81

7 6 6 4 4 3 4

288 228 228 128 183 109 163

0.66 0.41 1.63 2.13 0.46 0.69 0.65

0.32, 1.33 0.15, 1.14 0.55, 4.84 0.49, 9.24 0.13, 1.58 0.10, 4.54 0.15, 2.72

.24 .09 .38 .31 .22 .70 .55

.93 .70 .81 .92 .57 .53 .61

8 3

409 129

⫺2.60 ⫺0.49

⫺3.83, ⫺1.36 ⫺1.00, 0.01

<.001 .05

.18 .04

CI, Confidence interval; HG, heterogeneity; OR, odds ratio; WMD, weighted mean difference. *Statistically significant results are shown in bold.

211.e3 Simillis et al

Surgery February 2007

Fig 1. Test for heterogeneity: chi-square statistic with its degrees of freedom (df) and P value. Inconsistency among results: I2. test for overall effect: Z statistic with P value. CI, Confidence interval; OR, odds ratio; WMD, weighted mean difference.

Simillis et al 211.e4

Surgery Volume 141, Number 2

Table IV. Sensitivity analysis performed for studies comparing laparoscopic versus open surgery* Outcome Studies with size of laparoscopic group ⱖ 20 Operative time (min) Operative blood loss (mL) Overall postoperative complications Resection-related complications General complications Duration of hospital stay (d) High-quality studies ⱖ 6 stars Operative time (min) Operative blood loss (mL) Overall postoperative complications Resection related complications General complications Duration of hospital stay (d) Studies published after and including 2003 Operative time (min) Operative blood loss (mL) Overall postoperative complications Resection related complications General complications Duration of hospital stay (d) Studies matched for the presence of malignancy Operative time (min) Operative blood loss (mL) Overall postoperative complications Resection-related complications General complications Duration of hospital stay (d) Studies matched for segment resection Operative time (min) Operative blood loss (mL) Overall postoperative complications Resection-related complications General complications Duration of hospital stay (d)

No. of studies

No. of resections

OR/WMD

95% CI

P value

HG P value

2 3 3 2 2 4

100 160 160 100 100 281

⫺4.86 ⫺116.81 0.73 0.74 0.69 ⫺2.78

⫺52.81, 43.08 ⫺187.06, ⫺46.57 0.25, 2.09 0.11, 5.21 0.12, 3.85 ⫺4.67, ⫺0.89

.84 .001 .56 .76 .67 .004

.008 .47 .82 .25 .58 .05

4 5 5 4 4 5

152 212 212 152 152 212

28.69 ⫺148.68 0.55 0.24 1.88 ⫺3.24

⫺30.79, 88.18 ⫺217.31, ⫺80.06 0.25, 1.20 0.07, 0.80 0.55, 6.43 ⫺4.89, ⫺1.59

.34 <.001 .13 .02 .32 ⬍.001

<.001 .85 .96 .98 .78 .35

3 4 4 3 3 5

123 183 183 123 123 304

35.51 ⫺151.42 0.59 0.23 1.84 ⫺3.04

⫺34.75, 105.77 ⫺220.43, ⫺82.42 0.25, 1.40 0.06, 0.85 0.45, 7.54 ⫺4.54, ⫺1.54

.32 <.001 .23 .03 .4 <.001

<.001 .85 .93 .92 .58 .41

4 4 4 4 4 4

159 159 159 159 159 159

22.77 ⫺156.8 0.63 0.33 1.92 ⫺3.94

⫺29.70, 75.23 ⫺234.99, ⫺78.61 0.25, 1.61 0.08, 1.34 0.53, 7.00 ⫺6.08, ⫺1.80

.4 <.001 .33 .12 .32 <.001

<.001 .79 .74 .82 .77 .33

3 4 4 3 3 4

114 174 174 114 114 174

44.98 ⫺114.8 1.03 0.78 1.92 ⫺1.52

22.26, 67.69 ⫺191.38, ⫺38.23 0.36, 2.94 0.16, 3.85 0.30, 12.22 ⫺2.17, ⫺0.87

<.001 .003 .96 .76 .49 <.001

.16 .56 .92 .37 .4 .8

CI, Confidence interval; HG, heterogeneity; OR, odds ratio; WMD, weighted mean difference. *Statistically significant results are shown in bold.