Laparoscopic versus open liver resection for colorectal liver metastases: A systematic review and meta-analysis of studies with propensity score-based analysis

Accepted Manuscript Laparoscopic versus open liver resection for colorectal liver metastases: A systematic review and meta-analysis of studies with propensity score-based analysis Xue-Liang Zhang, Rui-Feng Liu, Dan Zhang, Yu-Sheng Zhang, Tao Wang PII:

S1743-9191(17)30458-2

DOI:

10.1016/j.ijsu.2017.05.073

Reference:

IJSU 3863

To appear in:

International Journal of Surgery

Received Date: 17 April 2017 Revised Date:

29 May 2017

Accepted Date: 31 May 2017

Please cite this article as: Zhang X-L, Liu R-F, Zhang D, Zhang Y-S, Wang T, Laparoscopic versus open liver resection for colorectal liver metastases: A systematic review and meta-analysis of studies with propensity score-based analysis, International Journal of Surgery (2017), doi: 10.1016/ j.ijsu.2017.05.073. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Laparoscopic versus open liver resection for colorectal liver metastases: a systematic review and meta-analysis of studies with propensity score-based analysis Xue-Liang Zhanga,b., Rui-Feng Liua,b*., Dan Zhangc., Yu-Sheng Zhanga., Tao Wangd

AC C

EP

TE D

M AN U

SC

RI PT

a The department of medical oncology in digestive system, The cancer hospital of Gansu Province, Lanzhou, China. b The first clinical medical college of Lanzhou University, Lanzhou, China. c Department of General surgery, The second clinical medical college of Lanzhou University, Lanzhou, China. d Translational medicine research center, Gansu provincial academic institute for medical research，China. *Corresponding to Rui-Feng Liu, Attending physician, PhD candidate in graduate students, specialize in radiotherapy and comprehensive treatment of cancer, worked in The cancer hospital of Gansu Province, Lanzhou, 730050, China.

ACCEPTED MANUSCRIPT

Laparoscopic versus open liver resection for colorectal liver metastases: a systematic review and meta-analysis of studies

RI PT

with propensity score-based analysis Abstract Background

SC

This meta-analysis collected studies with propensity score matching analysis (PSM) and focused on comparing the short-term and oncological outcomes of patients with

M AN U

colorectal liver metastases (CRLM) who underwent laparoscopic liver resection (LLR) versus open liver resection (OLR), to provide relatively high-level evidence of the additional value of LLR in treating patients with CRLM in comparison with OLR. Methods

A systematic literature search was performed using the PubMed, Embase and

TE D

Cochrane Library databases. Bibliographic citation management software (EndNote X7) was used for literature management. Quality assessment was performed based on a modified version of the Newcastle-Ottawa Scale. The data were analyzed using Review Manager (Version 5.1), and sensitivity analysis was performed by omitting

EP

one study in each step. Dichotomous data were calculated by odds ratio (OR) and continuous data were calculated by weighed mean difference (WMD) with 95%

AC C

confidence intervals (CI). Results

Overall, 10 studies enrolling 2259 patients with CRLM were included in the present meta-analysis. The pooled analysis suggested that LLR was associated with significantly less overall morbidity (OR, 0.57; 95% CI 0.40 to 0.80; I2= 57%; P <0.001), reduced blood loss (WMD, −124.68; 95% CI, −177.35 to −72.01; I2 =  83%; P <0.00001), less transfusion requirement (OR, 0.46; 95% CI 0.35 to 0.62; I2 =  0%; P <0.00001), shorter length of hospital stay (WMD, −2.13; 95% CI, −2.68 to −1.58; I2 =  0%; P <0.00001), but longer operative time (WMD, 39.48; 95% CI, 23.68 to 55.27; I2

ACCEPTED MANUSCRIPT  = 66%; P= 0.04). However, no significant differences were observed in mortality (OR, 0.50; 95% CI 0.21 to 1.2; I2= 0%; P= 0.12). For oncological outcomes, no significant differences were observed in negative surgical margins (R0 resection), tumor recurrence, 3-year disease-free survival, 5-year disease-free survival, 5-year

RI PT

overall survival between the approaches. Nevertheless, LLR tended to provide higher 3-overall survival rate (OR, 1.37; 95 % CI 1.11 to 1.69; I2= 0%; P= 0.003). The pooled OR for overall morbidity in each subgroup analysis was consistent with the overall pooled OR. Additionally, the pooled OR for overall morbidity varied from

39%; P <0.0001) in sensitivity analysis.

M AN U

Conclusion

SC

(0.63; 95% CI 0.45to 0.88; I2= 49%; P= 0.007) to (0.51; 95% CI 0.37 to 0.69; I2=

LLR is a beneficial alternative to OLR in select patients, and provides more favorable short-term outcomes such as less overall morbidity, shorter length of hospital stay, less blood loss, lower blood transfusion rate. Simultaneously, LLR does not compromise oncological outcomes including surgical margins R0, tumor recurrence,

TE D

disease-free survival, 5-overall survival, as well as even yielding better 3-overall survival. Considering unavoidable bias from non-randomized trials, high-quality RCTs are badly needed to determine whether LLR can become standard practice for

EP

treating patients with CRLM.

Introduction

AC C

Colorectal cancer is the second most common cause in Western Europe and America, and 15%–25% patients with colorectal cancer have synchronous liver metastases (1, 2). Colorectal liver metastases (CRLM) has been widely recognized as an absolute indication for liver resection and the complete resection of primary colorectal cancer (CRC) and CRLM still offers the greatest curative potential, although there are many choices including ablative therapy, cryotherapy, perioperative chemotherapy, and hepatic artery infusion (3, 4). In recent years, a large body of literature has highlighted the added values of laparoscopic liver resection (LLR) versus open liver resection (OLR) in terms of reduced blood loss, lower morbidity rate, shorter length of hospital

ACCEPTED MANUSCRIPT stay and earlier return to functional activities (5-12). Furthermore, LLR was even thought to provide superior oncological outcomes to patients with CRLM (13). However, the positive results from LLR have to be interpreted discreetly due to the potential role of selection bias and small sample size that may lead to inauthentic

RI PT

evaluation of outcomes and uncertain conclusions in that the majority of relevant data were merely extracted from case series, case-control studies or meta-analyses of these studies(14-21). Consistent with the perceived benefits from the studies above, in 2014, the Second International Consensus Conference on LLR (ICCLLR) in 2014

SC

concluded that minor LLR had become standard practice and that major liver resections were still innovative procedures in the exploration phase that should be

M AN U

practiced with continued caution. However, the only low-level evidence available for scrutiny, and higher-quality evaluative studies are badly needed to determine the place of LLR in treating liver lesions (22). Currently, two ongoing clinical trials, the Oslo CoMet trial (NCT01516710) and the multinational Orange II Plus trial (NCT01441856), have randomized patients with resectable CRLM to LLR or OLR,

TE D

but the final results of both trials have not yet been reported. However, treading on the heels of the second ICCLLR and lacking data from RCTs, the researchers in this field have conducted numerous high-quality non-randomized trials based on propensity

EP

score matching (PSM) analyses to assess the additional value of LLR in patients with CRLM compared with OLR. The propensity score (PS) method is a successful tool for minimizing treatment selection bias in the context of observational studies and the

AC C

recent literature even indicated that non-randomized studies with suitable PSM analysis could provide comparable evidence to that from randomized controlled trials (RCT) (23). In these PSM analysis-dependent studies comparing the outcomes of CRLM patients undergoing LLR versus OLR, propensity score adjustment was performed on variables such as baseline characteristics, tumor characteristics, preoperative chemotherapy and type of liver resection, which are known to influence the choice of treatment approach and prognosis. To date, most of the studies with propensity score matching analysis suggested that LLR for CRLM may provide oncological outcomes comparable to those from OLR, as well as better short-term

ACCEPTED MANUSCRIPT outcomes including less blood loss, lower blood transfusion rate, lower overall morbidity, and shorter hospital stay in selected patients(24-27). Conversely, some of these studies indicated that there was no significant difference regarding incidence of morbidity and mortality (13, 28) and that LLR have even resulted in improved

RI PT

oncological outcomes (13). The sample size from each trial ranged from 36 to 450, which indicated that small sample issues may partly explain the inconsistencies across different trials and influence the confidence level of the evidence from these studies via PSM analysis (13, 24-27, 29-33). The most recently published meta-analysis

SC

comparing the effects of LLR and OLR in CRLM patients was conducted by Tian ZQ et al, but the meta-analysis only included two studies with PSM, which mean that the

M AN U

quality of evidence was rather low-level (19). Therefore, this meta-analysis of high-quality studies with PSM analysis is indispensable and was conducted to , provide the current best evidence supporting LLR in treating patients with CRLM.

Methods

The current meta-analysis was undertaken according to the Cochrane Handbook for

TE D

Systematic Reviews of Interventions and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (34, 35). Literature search

EP

The PubMed, Embase, and Cochrane Library were searched from inception until Apr. 2017. The electronic searches were performed using exploded medical subject headings (MeSH) and the corresponding terms, including “laparoscop*”, “minimally

AC C

invasive”, “open’, “liver metasta*”, “hepatic metasta*”, “colorectal metasta*”, “colon metasta*”, “rectal metasta*”, and “rectum metasta*”. In addition, the references of relevant reviews and included studies were checked to retrieve any other possible eligible studies.

Study selection We selected all clinical studies comparing short-term and oncological results of patients with CRLM who underwent laparoscopic and open liver resection. Additionally, all the studies included in the current meta-analysis had to fulfill the criteria: (1) designed with PSM analysis (2) published as full-length articles.

ACCEPTED MANUSCRIPT Furthermore, the following studies were also excluded: (1) abstracts, letters, protocol, comments, reviews, non-matched studies and matched studies without PSM analysis. (2) studies that lack clearly reported outcomes of interest; (3) studies that evaluated patients with benign lesions or malignant non-CRLM. Data Extraction and Quality Assessment

RI PT

The following parameters were extracted from each study and tabulated by two reviewers (Xue-liang Zhang and Dan Zhang): first author names, year of publication, study design, number of patients in each group, patient baseline characteristics, and

SC

outcomes of interest. The primary outcome was overall postoperative morbidity. The secondary outcomes were operative time, blood loss, blood transfusion requirement,

M AN U

length of hospital stay, 90-day mortality and oncological outcomes including surgical margins R0, tumor recurrence, 3-year disease-free survival, 5-year disease-free survival, 3-year overall survival and 5-year overall survival. All relevant text, tables and figures were reviewed for data extraction. Discrepancies between the reviewers were resolved by discussion. The methodological quality of the included studies was

TE D

evaluated in accordance with the Newcastle-Ottawa Scale (NOS), which scores patient selection, study group comparability, and outcome assessment (36). Studies with more than 6 points (maximum 11) were defined as high quality. Statistical Methods

EP

The meta-analysis was conducted using Review Manager (RevMan) and Stata 12.0 software, version 5.2. Considering foreseeable clinical heterogeneity of participants

AC C

and interventions, this meta-analysis was performed based on random-effects model, which assumes that studies do not share the identical common effect and distributes a weight to each study, considering both within- and between-study variances. We analyzed dichotomous variables using estimation of odds ratio (OR) with a 95% confidence interval (95% CI) and continuous variables using weighted mean difference (WMD) with a 95% CI. The heterogeneity among the included studies was estimated with I2 values; I2 >50% was considered significant heterogeneity. To further investigate sources of heterogeneity and to confirm the stability of pooled estimation of the primary outcome in different subsets, we performed subgroup

ACCEPTED MANUSCRIPT analyses and sensitivity analysis by omitting one study at each step to determine the influence of a single study on the overall pooled estimate. The results were considered statistically significant at two-sided p-values<0.05. The risk of publication bias was assessed by funnel plots and Begg and Egger test (37, 38).

RI PT

Results Study selection and Study Characteristics

Figure 1 illustrates the study screening and selection process. We identified 1740 studies from three databases (PubMed, Embase and The Cochrane Library). After

SC

removing duplicates and screening the titles and abstracts, the remaining 70 were considered for potential eligibility. Then, we further reviewed these full texts; Finally

meta-analysis (13, 24-27, 29-33).

M AN U

10 publications of non-randomized trials with PSM analysis were included in present

Table 1 summarizes the characteristics of the included studies and details of the enrolled participants. The sample size of included studies ranged from 36 to 450. A total of 2259 patients (LLR 980 and OLR 1279) from ten non-randomized trials with

TE D

PSM analysis were included. Among the included studies, two were performed only in elderly patients (32, 33) and five trials were conducted by single-center (13, 25-27, 33). In addition, the patients in two of the included studies underwent liver resection

EP

with synchronous colectomy. Across all the non-randomized trials with PSM analysis, the percentage of major hepatectomy ranged from 0% to 55.8% and three of them

AC C

involved> 40% major hepatectomy in both groups (24-26). Assessment of study quality The risks of bias were assessed using a modified Newcastle–Ottawa scale (Table 2). Full-length articles were all available for review. Five trials were conducted by multiple institutions. The liver resections occurred between 1997 and 2016, which indicated a significant time bias. The sample size of included studies ranged from 36 to 450; therefore, the small sample size may have influenced the authenticity. Considering ethics, specificity and personal preference, surgeons can hardly randomly allocate CRLM patients into laparoscopic or open liver resection. Thus, selection bias is unavoidable to some degree, although PSM analysis can minimize this issue.

ACCEPTED MANUSCRIPT Exposure ascertainment and postoperative assessment were confirmed using surgical records.

The PSM analysis was performed based on demographic data, American

Society of Anesthesiology score, tumor size, type of resection, tumor number, number of resected segments, tumor location, and minor resections. The patient outcomes may

RI PT

have been influenced by selection bias. The methods for handling missing were not adequately described in most of the studies. Overall, all studies were evaluated as being moderate to high quality.

Meta-analysis of short-term outcomes

SC

With respect to surgical outcomes, six endpoints including operation time, perioperative blood loss, blood transfusion requirement, postoperative overall

M AN U

morbidity, postoperative 90-day mortality, and hospitalization time were considered. The dichotomous data were assessed data was calculated by OR, and the continuous data was assessed by WMD with 95% CI. Operation time

The operation time was available from 8 studies (13, 26, 27, 29-33).The pooled

TE D

analysis indicated that LLR was associated with longer operative time, but there was significant heterogeneity among the studies (WMD, 39.48; 95%CI, 23.68 to 55.27; I2= 66%; P= 0.04) (Figure 2 A).

EP

Blood loss

The mean changes in blood loss were pooled for the 9 studies (13, 25-27, 29-33). The

AC C

pooled result showed that LLR was associated with less blood loss (WMD, −124.68; 95%CI, −177.35 to −72.01; I2= 83%;P <0.00001) than was OLR (Figure 2 B). Blood transfusion

Eight studies reported perioperative blood transfusion (13, 24, 25, 27, 29-32). The pooled analysis indicated that the rate of patients requiring blood transfusion was lower in LLR than in OLR (OR, 0.46; 95% CI 0.35 to 0.62; I2= 0%; P <0.00001) (Figure 2 C). Overall morbidity Ten studies mentioned the postoperative morbidity rate (13, 24-27, 29-33). The results

ACCEPTED MANUSCRIPT showed that patients in the LLR group suffered lower morbidity than in the OLR group (OR, 0.57; 95% CI 0.40 to 0.80; I2= 57%; P <0.001) (Figure 3 A). Ninety-day mortality The relative risk of postoperative mortality was mentioned in 7 studies (24-27, 29, 30,

RI PT

32). The pooled estimate for postoperative 90-day mortality indicated there was no significant difference between LLR and OLR (OR, 0.50; 95% CI 0.21 to 1.2; I2= 0%; P = 0.12) (Figure 3 B). Hospitalization time

SC

The length of hospital stay was pooled for 9 studies (13, 24, 26, 27, 29-33). From the pooled analysis, shorter hospitalization time was observed in LLR group when

<0.00001) (Figure 3 C).

M AN U

compared with OLR group (WMD, −2.13; 95% CI, −2.68 to −1.58; I2= 0%; P

Meta-analysis of oncological outcomes

Six oncological outcomes including negative surgical margins (R0 resection), tumor recurrence, 3-year disease-free survival, 5-year disease-free survival, 5-year overall

TE D

survival and 5-year overall survival were analyzed. R0 resection

The R0 resection was reported in 8 studies (24, 26, 27, 29-32).The pooled analysis

EP

showed that surgical margin R0 was similar between LLR and OLR (OR, 1.23; 95% CI 0.84 to 1.8; I2= 25%; P = 0.29) (Figure 4 A).

AC C

Recurrence

Tumor recurrence combining hepatic and extra-hepatic recurrence was mentioned in 6 studies (13, 24, 26, 27, 29, 30). The pooled estimate for tumor recurrence suggested LLR was comparable to OLR (OR, 0.93; 95% CI 0.66 to 1.3; I2= 47%; P= 0.66) (Figure 4 B). Disease-free survival The 3-year disease-free survival was available for all the included studies (13, 24-27, 29-33) and 8 studies reported the 5-year disease-free survival (13, 24-27, 29, 32, 33). The pooled analysis indicated that LLR provided comparable 3-year disease-free

ACCEPTED MANUSCRIPT survival (OR, 0.91; 95% CI, 0.61 to 1.38; I2= 78%; P= 0.67) (Figure 4 C) and 5-year disease-free survival (OR, 0.95; 95% CI, 0.68 to 1.32; I2= 56%; P= 0.76) (Figure 5 A). Overall survival

RI PT

Ten studies reported the 3-year overall survival (13, 24-27, 29-33), and 8 studies described the 5-year overall survival (13, 24-27, 29, 32, 33). The pooled OR for the 5-year overall survival was parallel in LLR and OLR groups (OR, 1.05; 95% CI, 0.86 to 1.27; I2= 0%; P= 0.65). (Figure 5 B). However, the pooled analysis indicated that

CI, 1.11 to 1.69; I2= 0%; P= 0.003) (Figure 5 C).

M AN U

Subgroup and sensitivity analyses

SC

LLR was associated with better 3-year overall survival than was OLR (OR, 1.37; 95%

Subgroup analyses were performed to explore the source of heterogeneity and confirm whether the pooled estimate of the primary endpoint, the overall postoperative morbidity, was stable according to clinical setting (single-center or multi-center), sample size (>300 or <300), type of resection (simul or liver/simul),

TE D

percentage of major hepatectomy (>40% or <40%) and age (only elderly or mixed-age patients). The results indicated that the uneven percentage of major hepatectomy across all the include studies may have be partially responsible for the

EP

observed heterogeneity. Because, when all ten studies were stratified into two subgroups according to the percentage of major hepatectomy (>40% or <40%), the I2

AC C

(57%) for all the included studies declined to 0% and 29% respectively. However, the other four issues did not contribute to the study heterogeneity. Additionally, the pooled OR in each subgroup analysis supported that LLR was associated lower morbidity compared with OLR, as the pooled OR from all the included studies did. The detailed results are presented in (Table 3). Furthermore, sensitivity analyses were performed by omitting one study at each step to determine the influence of a single study on the overall pooled estimate. The results showed the pooled OR for overall morbidity varied from (0.63; 95 % CI 0.45 to 0.88; I2= 49%; P= 0.007) to (0.51; 95 % CI 0.37 to 0.69; I2= 39%; P <0.0001), which indicated that the pooled OR for overall

ACCEPTED MANUSCRIPT morbidity was stable. Publications bias There was no obvious publication bias after visual inspection of the funnel plot and binary variable statistical analysis (overall morbidity: Egger test, p=0.959; Begg test, p=0.858) (Figure 6).

RI PT

Discussion Main findings

Ten non-randomized trials with PSM analysis that compared the effects of LLR

SC

versus OLR on the short-term and oncological outcomes in patients with CRLM were identified. The present meta-analysis suggested that LLR is a beneficial alternative to

M AN U

OLR in select patients and provides more favorable short-term outcomes such as less overall morbidity, shorter length of hospital stay, less blood loss, lower blood transfusion rate. However, LLR does not compromise oncological outcomes including surgical margin R0, tumor recurrence, disease-free survival, 5-overall survival. Furthermore, LLR even yield better 3-overall survival compared with OLR than does OLR.

TE D

Comparison with other meta-analysis

Five meta-analyses of observational studies on this topic have been performed (14, 16, 19-21). There are two notable points in the present meta-analysis compared with the

EP

previous meta-analyses. First, the present meta-analysis provided new evidence supporting that LLR is superior to OLR in treating patients with CRLM. On the one

AC C

hand, the present meta-analysis indicated that LLR was associated with better 3-overall survival, which was not reported in the previous meta-analyses. On the other hand, the subgroup analysis according to age (only elderly or mixed-age patients) indicated that elderly patients could experience additional benefit from LLR in terms of lower morbidity. In addition, the subgroup analysis performed according to the percentage of major hepatectomy (>40% or <40%) reinforced the superiority of LLR over OLR and indicated that LLR was also a potentially safe alternative to OLR in treating patients who needed major hepatectomy. These results may be a positive response to the conclusion made by the Second International Consensus Conference

ACCEPTED MANUSCRIPT that major liver resections were still innovative procedures during the exploration phase and should be practiced with continued caution. Both of these aspects mentioned above have not been reported in previous meta-analyses. Second, the present meta-analysis only included non-randomized trials with PSM analyses, which

RI PT

could minimize the treatment selection bias. Furthermore, PSM analyses could even provide comparable evidence to RCTs if the former are performed properly based on variables such as baseline characteristics, tumor characteristics, preoperative chemotherapy and type of liver resection which are known to influence the choice of

SC

treatment approach. However, the majority of studies included in the prior meta-analyses were observational studies without PSM analyses, which were

M AN U

challenged by significant bias and might incur overestimated pooled effects. Therefore, the evidence that the present review provided might be the best current evidence supporting that LLR provides the added benefits for select patients with CRLM compared with OLR. Implications for clinical practice

TE D

The present meta-analysis showed that LLR was related to longer operative time, which might be explained by the intricacy of LLR procedure. Despite the idea that LLR is challenged by difficult-to-control blood loss due to its indirect and intricate manipulation, the present meta-analysis indicated that blood loss and transfusion

EP

requirement in LLR group were significantly lower than those in OLR group. The differences in blood loss and transfusion requirement may be interpreted by the

AC C

magnified images and more meticulous dissection, pneumoperitoneum, and less blood loss from the abdominal wall provided by laparoscopic approach (39). The volume of blood loss and the requirement for blood transfusion were closely related to increased postoperative morbidity (40). Consistent with this observation, the present meta-analysis also indicated that LLR had a clear advantage over OLR in reducing postoperative morbidity and hospitalization time. Furthermore, subgroup analysis confirmed our finding regarding overall morbidity. Usually, the cost of laparoscopic equipment is relatively higher than that of open counterpart, which increases the economic burden for patients undergoing LLR for CRLM. However, the benefits of

ACCEPTED MANUSCRIPT reduced blood loss, shorter hospitalization time and lower morbidity from LLR might offset the higher costs of the laparoscopic equipment. Although we did not compare the costs between the two procedures owing to limited data, a comparison of laparoscopic versus open left lateral sectionectomy in a deviation based cost modeling

RI PT

study indicated that the cost of each patient undergoing LLR was significantly less than that for each patient who underwent OLR (41). Anyway, more efforts should be devoted to comparing the economic effects of the two procedures in future studies to understand the potential wide application of LLR in patients with CRLM.

SC

The present meta-analysis indicated that LLR was similar to OLR in the oncological outcomes such as surgical margins R0, tumor recurrence, 3-year disease-free survival,

M AN U

5-year disease-free survival and 5-year overall survival, which was consistent with the previous meta-analysis. LLR is an intricate surgical procedure in which it is relatively more difficult for surgeons to achieve R0 resection than with OLR. However, intraoperative ultrasonography-guided technique has been widely used to assist liver surgeons in precisely locating the tumor-free margin precisely (42, 43), and this may

TE D

partially explain the similar pooled effects of R0 resection in two groups in this meta-analysis. With respect to the similar oncological outcomes between LLR and OLR, this effect could be partly interpreted by the changes to auxiliary therapy during

EP

the research. The included studies were completed between 1997 and 2016, during which chemotherapy and targeted therapy advanced immensely. Thus, the differences in chemotherapy and targeted therapy may offset the prognostic values of LLR in

AC C

CRLM management as a confounding factor if the number of patients in the early period in LLR group were more than those in OLR. Actually, it remains debatable whether LLR was able to provide better oncological outcomes to patients with CRLM than was OLR. For example, the present meta-analysis indicated that LLR was associated with a higher 3-year overall survival. In addition, LRR provided a lower rate of tumor recurrence and survival benefit than previously reported (13, 39). The delayed recovery due to postoperative morbidity might enhance the immune response impairment and postpone adjuvant treatments following surgery, both of which accelerate relapses and shorten long-term survival. Thus, the less postoperative

ACCEPTED MANUSCRIPT morbidity may be partially responsible for LLR-related improved oncological outcomes. Additionally, two conference abstracts of randomized clinical trials reported that LLR was related to a lesser inflammatory response in CRLM patients with a reduced release of oncogenic proteins as high-mobility group box 1 (HMGB-1)

RI PT

(44, 45), which might also support the survival benefit of LLR in treating patients with CRLM. Regardless, high quality trials are anticipated to determine whether LLR is able to provide better oncological outcomes to patients with CRLM compared to OLR.

SC

Limitations and conclusions

There are several caveats in the present meta-analysis that affect the interpretation of

M AN U

the results. First, all the included studies were non-randomized trials. Despite the ability of PSM analysis to minimize the selection bias, any non-randomized trial is unavoidably challenged by undetected bias, which affects the reliability of the pooled effects. Second, although the subgroup analyses and sensitivity analyses proved the robustness of pooled effect estimates for overall morbidity, considerable

TE D

heterogeneity across the studies cannot be overlooked in our meta-analysis owing to the inherent differences from baseline data and operation experience. For example, the percentage of major hepatectomy differed significantly across all the included trials, which might directly affect the application of the conclusion obtained from the

EP

present meta-analysis to CRLM patients who need major liver resection. Third, in the subgroup analysis by age (only elderly or mixed-age patients), only two studies

AC C

enrolled 608 patients had available data comparing the outcomes between LLR and OLR in elderly patients with CRLM, suggesting the selection bias related to small sample issue may result in overestimated pooled effects of outcomes. Therefore, more high-quality and high-volume trials are needed to determine whether LLR is associated better outcomes in CRLM elderly patients. Fourth, the patients undergoing LLR in the majority of the included studies were selected by the current indications for laparoscopic procedure including tumor location, tumor size, and number of liver lesion, so it is difficult to generalize our findings to all CRLM patients. Last but not least, potential advantages of laparoscopy including pain control, recovery time,

ACCEPTED MANUSCRIPT esthetics and quality of life, were not assessed in the present systematic review due to limited data. In spite of aforementioned limitations, the present meta-analysis revealed the best current evidence reinforcing the theory that LLR is a beneficial alternative to OLR in

RI PT

select patients and provides more favorable short-term outcomes such as less overall morbidity, shorter length of hospital stay, less blood loss, lower blood transfusion rate. At the same time, LRR does not compromise oncological outcomes including surgical margin R0, tumor recurrence, disease-free survival, 5-overall survival. Furthermore,

SC

LRR even yields a better 3-overall survival. Considering unavoidable bias from non-randomized trials, high-quality RCTs are badly needed to determine whether

M AN U

LLR can become standard practice for treating patients with CRLM. FIG. 1 illustrates the study screening and selection process.

FIG. 2 Forest plots depicting operative time, blood loss and blood transfusion in the

TE D

included studies. ORs and WMDs are shown with 95% CIs.

FIG. 3 Forest plots depicting overall morbidity, 90-day mortality and hospitalization time in the included studies. ORs and WMDs are shown with 95% CIs.

EP

FIG.4 Forest plots depicting surgical margins R0, recurrence and 3-year DFS in the

AC C

included studies. ORs and WMDs are shown with 95% CIs.

FIG. 5 Forest plots depicting 3-year DFS, 3-year OS and 3-year OS reported in the included studies. ORs are shown with 95% CIs

FIG. 6 Funnel plots depicting the publication bias in the included studies

Acknowledgment This study was not supported by any pharmaceutical company or grants; the cost was borne by the authors’ institutions. The authors thank “American Journal Experts” who

ACCEPTED MANUSCRIPT provided support for amending language of this article. Neither of these persons received compensation for the work performed.

Reference

Leporrier J, Maurel J, Chiche L, Bara S, Segol P, Launoy G. A population-based study of the

RI PT

1.

incidence, management and prognosis of hepatic metastases from colorectal cancer. Br J Surg. 2006;93(4):465-74. Epub 2006/03/09. 2.

Siegel R, Naishadham D, Jemal A. Cancer statistics, 2013. CA Cancer J Clin. 2013;63(1):11-30.

Epub 2013/01/22.

Choti MA, Sitzmann JV, Tiburi MF, Sumetchotimetha W, Rangsin R, Schulick RD, et al. Trends

SC

3.

in long-term survival following liver resection for hepatic colorectal metastases. Ann Surg. 2002;235(6):759-66. Epub 2002/05/30. 4.

Fong Y, Fortner J, Sun RL, Brennan MF, Blumgart LH. Clinical score for predicting recurrence

M AN U

after hepatic resection for metastatic colorectal cancer: analysis of 1001 consecutive cases. Ann Surg. 1999;230(3):309-18; discussion 18-21. Epub 1999/09/24. 5.

Castaing D, Vibert E, Ricca L, Azoulay D, Adam R, Gayet B. Oncologic results of laparoscopic

versus open hepatectomy for colorectal liver metastases in two specialized centers. Ann Surg. 2009;250(5):849-55. Epub 2009/10/06. 6.

Cheung TT, Poon RT, Yuen WK, Chok KS, Tsang SH, Yau T, et al. Outcome of laparoscopic

versus open hepatectomy for colorectal liver metastases. ANZ J Surg. 2013;83(11):847-52. Epub 7.

TE D

2012/10/06.

Hasegawa Y, Nitta H, Sasaki A, Takahara T, Itabashi H, Katagiri H, et al. Long-term outcomes of

laparoscopic versus open liver resection for liver metastases from colorectal cancer: A comparative analysis of 168 consecutive cases at a single center. Surgery. 2015;157(6):1065-72. Epub 2015/03/21. 8.

Karagkounis G, Akyuz M, Guerron AD, Yazici P, Aucejo FN, Quintini C, et al. Perioperative and

EP

oncologic outcomes of minimally invasive liver resection for colorectal metastases: A case-control study of 130 patients. Surgery. 2016;160(4):1097-103. Epub 2016/08/04. 9.

Qiu J, Chen S, Pankaj P, Wu H. Laparoscopic hepatectomy is associated with considerably less

AC C

morbidity and a long-term survival similar to that of the open procedure in patients with hepatic colorectal metastases. Surg Laparosc Endosc Percutan Tech. 2014;24(6):517-22. Epub 2014/09/05. 10. Schiffman SC, Kim KH, Tsung A, Marsh JW, Geller DA. Laparoscopic versus open liver resection for metastatic colorectal cancer: a metaanalysis of 610 patients. Surgery. 2015;157(2):211-22. Epub 2014/10/06.

11. Tohme S, Goswami J, Han K, Chidi AP, Geller DA, Reddy S, et al. Minimally Invasive Resection of Colorectal Cancer Liver Metastases Leads to an Earlier Initiation of Chemotherapy Compared to Open Surgery. J Gastrointest Surg. 2015;19(12):2199-206. Epub 2015/10/07. 12. Topal H, Tiek J, Aerts R, Topal B. Outcome of laparoscopic major liver resection for colorectal metastases. Surg Endosc. 2012;26(9):2451-5. Epub 2012/02/24. 13. Ratti F, Catena M, Di Palo S, Staudacher C, Aldrighetti L. Impact of totally laparoscopic combined management of colorectal cancer with synchronous hepatic metastases on severity of complications: a propensity-score-based analysis. Surg Endosc. 2016;30(11):4934-45. Epub

ACCEPTED MANUSCRIPT 2016/10/28. 14. Hallet J, Beyfuss K, Memeo R, Karanicolas PJ, Marescaux J, Pessaux P. Short and long-term outcomes of laparoscopic compared to open liver resection for colorectal liver metastases. Hepatobiliary Surg Nutr. 2016;5(4):300-10. Epub 2016/08/09. 15. Langella S, Russolillo N, D'Eletto M, Forchino F, Lo Tesoriere R, Ferrero A. Oncological safety of ultrasound-guided laparoscopic liver resection for colorectal metastases: a case-control study. Updates Surg. 2015;67(2):147-55. Epub 2015/07/30.

RI PT

16. Luo LX, Yu ZY, Bai YN. Laparoscopic hepatectomy for liver metastases from colorectal cancer: a meta-analysis. J Laparoendosc Adv Surg Tech A. 2014;24(4):213-22. Epub 2014/02/28.

17. Montalti R, Berardi G, Laurent S, Sebastiani S, Ferdinande L, Libbrecht LJ, et al. Laparoscopic liver resection compared to open approach in patients with colorectal liver metastases improves further resectability: Oncological outcomes of a case-control matched-pairs analysis. Eur J Surg Oncol.

SC

2014;40(5):536-44. Epub 2014/02/22.

18. Nachmany I, Pencovich N, Zohar N, Ben-Yehuda A, Binyamin C, Goykhman Y, et al. Laparoscopic versus open liver resection for metastatic colorectal cancer. Eur J Surg Oncol. 2015;41(12):1615-20. Epub 2015/10/12.

M AN U

19. Tian ZQ, Su XF, Lin ZY, Wu MC, Wei LX, He J. Meta-analysis of laparoscopic versus open liver resection for colorectal liver metastases. Oncotarget. 2016. Epub 2016/11/05. 20. Wei M, He Y, Wang J, Chen N, Zhou Z, Wang Z. Laparoscopic versus open hepatectomy with or without synchronous colectomy for colorectal liver metastasis: a meta-analysis. PLoS One. 2014;9(1):e87461. Epub 2014/02/04.

21. Zhou Y, Xiao Y, Wu L, Li B, Li H. Laparoscopic liver resection as a safe and efficacious alternative to open resection for colorectal liver metastasis: a meta-analysis. BMC surgery. 2013;13:44.

TE D

22. Wakabayashi G, Cherqui D Fau - Geller DA, Geller Da Fau - Buell JF, Buell Jf Fau - Kaneko H, Kaneko H Fau - Han HS, Han Hs Fau - Asbun H, et al. Recommendations for laparoscopic liver resection: a report from the second international consensus conference held in Morioka. (1528-1140 (Electronic)).

23. Lonjon G, Boutron I, Trinquart L, Ahmad N, Aim F, Nizard R, et al. Comparison of treatment

EP

effect estimates from prospective nonrandomized studies with propensity score analysis and randomized controlled trials of surgical procedures. Ann Surg. 2014;259(1):18-25. Epub 2013/10/08. 24. Allard MA, Cunha AS, Gayet B, Adam R, Goere D, Bachellier P, et al. Early and Long-term

AC C

Oncological Outcomes After Laparoscopic Resection for Colorectal Liver Metastases: A Propensity Score-based Analysis. Ann Surg. 2015;262(5):794-802. Epub 2015/11/20. 25. Cannon RM, Scoggins CR, Callender GG, McMasters KM, Martin RC, 2nd. Laparoscopic versus open resection of hepatic colorectal metastases. Surgery. 2012;152(4):567-73; discussion 73-4. Epub 2012/09/05.

26. Cipriani F, Rawashdeh M, Stanton L, Armstrong T, Takhar A, Pearce NW, et al. Propensity score-based analysis of outcomes of laparoscopic versus open liver resection for colorectal metastases. Br J Surg. 2016;103(11):1504-12. Epub 2016/08/04. 27. de'Angelis N, Eshkenazy R, Brunetti F, Valente R, Costa M, Disabato M, et al. Laparoscopic versus open resection for colorectal liver metastases: a single-center study with propensity score analysis. J Laparoendosc Adv Surg Tech A. 2015;25(1):12-20. Epub 2014/11/18. 28. Beppu T, Sakamoto Y, Hasegawa K, Honda G, Tanaka K, Kotera Y, et al. A nomogram predicting disease-free survival in patients with colorectal liver metastases treated with hepatic

ACCEPTED MANUSCRIPT resection: multicenter data collection as a Project Study for Hepatic Surgery of the Japanese Society of Hepato-Biliary-Pancreatic Surgery. J Hepatobiliary Pancreat Sci. 2012;19(1):72-84. Epub 2011/10/25. 29. Beppu T, Wakabayashi G, Hasegawa K, Gotohda N, Mizuguchi T, Takahashi Y, et al. Long-term and perioperative outcomes of laparoscopic versus open liver resection for colorectal liver metastases with propensity score matching: a multi-institutional Japanese study. J Hepatobiliary Pancreat Sci. 2015;22(10):711-20. Epub 2015/04/24. 30. Tranchart H, Fuks D, Vigano L, Ferretti S, Paye F, Wakabayashi G, et al. Laparoscopic

RI PT

simultaneous resection of colorectal primary tumor and liver metastases: a propensity score matching analysis. Surg Endosc. 2016;30(5):1853-62. Epub 2015/08/16.

31. Untereiner X, Cagniet A, Memeo R, Tzedakis S, Piardi T, Severac F, et al. Laparoscopic hepatectomy versus open hepatectomy for colorectal cancer liver metastases: comparative study with propensity score matching. Hepatobiliary Surg Nutr. 2016;5(4):290-9. Epub 2016/08/09.

SC

32. Martinez-Cecilia D, Cipriani F Fau - Vishal S, Vishal S Fau - Ratti F, Ratti F Fau - Tranchart H, Tranchart H Fau - Barkhatov L, Barkhatov L Fau - Tomassini F, et al. Laparoscopic Versus Open Liver Resection For Colorectal Metastases in Elderly and Octogenarian Patients: A Multicenter Propensity Score Based Analysis of Short- and Long-term Outcomes. LID - 10.1097/SLA.0000000000002147

M AN U

[doi]. (1528-1140 (Electronic)).

33. Zeng Y, Tian M. Laparoscopic versus open hepatectomy for elderly patients with liver metastases from colorectal cancer. (1107-0625 (Print)).

34. Higgins J, Green S. Cochrane handbook for systematic reviews of interventions 5 version 5.1.0. The Cochrane Collaboration. 2011:Available from: http://www.cochrane.org/handbook. 35. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Bmj. 2009;339:b2535. Epub 2009/07/23. for

assessing

the

TE D

36. Wells GASB, O’Connell D, Peterson J, Welch V, Losos M. The Newcastle-Ottawa Scale (NOS) quality

of

nonrandomised

studies

in

meta-analyses.

.

2011:http://wwwohrica/programs/ clinical_epidemiology/oxfordasp. 37. Begg CB, Mazumdar M. Operating characteristics of a rank correlation test for publication bias. Biometrics. 1994;50(4):1088-101. Epub 1994/12/01.

EP

38. 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. Epub 1997/10/06. 39. Kazaryan AM, Marangos IP, Røsok BI, Rosseland AR, Villanger O, Fosse E, et al. Laparoscopic

AC C

resection of colorectal liver metastases: Surgical and long-term oncologic outcome. Annals of Surgery. 2010;252(6):1005-12.

40. Wu WC, Smith TS, Henderson WG, Eaton CB, Poses RM, Uttley G, et al. Operative blood loss, blood transfusion, and 30-day mortality in older patients after major noncardiac surgery. Ann Surg. 2010;252(1):11-7. Epub 2010/05/28. 41. Vanounou T, Steel JL, Nguyen KT, Tsung A, Marsh JW, Geller DA, et al. Comparing the clinical and

economic

impact

of

laparoscopic

versus

open

liver

resection.

Ann

Surg

Oncol.

2010;17(4):998-1009. Epub 2009/12/25. 42. Gigot JF, Glineur D, Santiago Azagra J, Goergen M, Ceuterick M, Morino M, et al. Laparoscopic liver resection for malignant liver tumors: preliminary results of a multicenter European study. Ann Surg. 2002;236(1):90-7. Epub 2002/07/20. 43. Uchiyama K, Ueno M, Ozawa S, Kiriyama S, Shigekawa Y, Yamaue H. Combined use of contrast-enhanced intraoperative ultrasonography and a fluorescence navigation system for identifying

ACCEPTED MANUSCRIPT hepatic metastases. World journal of surgery. 2010;34(12):2953-9. 44. Fretland ÅA, Kazaryan AM, Bjørnbeth BA, Røsok B, Flatmark K, Andersen MH, et al. Open versus laparoscopic liver resection of colorectal metastases: The OSLO-CoMET study. HPB. 2016;18:e59-e60. 45. Fretland AA, Sokolov A, Postriganova N, Kazaryan AM, Pischke SE, Nilsson PH, et al. Inflammatory Response After Laparoscopic Versus Open Resection of Colorectal Liver Metastases:

AC C

EP

TE D

M AN U

SC

RI PT

Data From the Oslo-CoMet Trial. Medicine (Baltimore). 2015;94(42):e1786. Epub 2015/10/27.

ACCEPTED MANUSCRIPT

country

Allard et al, 2015

France

Study duration 2006-2013

Study Clinical design setting

Type of resection

Arms LLR

R

Multi-center

liver/Simul

US

al, UK

de’Angelis et al, France 2014

2004-2010

2004-2015

2000-2013

R

R

Multi-center

Single-center

Single-center

R

Single-center

153

153

Age (years) ＞70 (n= 59) ＞70 (n= 57) ＞70 (n= 61) ＞70 (n= 103)

MH (n,％) 70 (45.7)

NA

NA

67 (43.7)

NA

NA

>5 (n= 4) >5 (n= 8)

>2 (n= 54) >2 (n= 91) median= 1 median= 1 Multiple (n= 79) Multiple (n= 77)

NA

Tumor size (cm)

LLR

171

OLR

342

LLR

35

62±11

19 (54)

mean =8

OLR

138

62±11

77 (55.8)

mean =6

LLR

133

65 (48.9)

OLR

133

＞70 (n= 48) ＞70 (n= 53)

LLR

52

63±12

18 (34.6)

OLR

52

63±9

22 (42.3)

liver/Simul

NA

liver/Simul

TE D

et

al,

R

EP

Cipriani 2016

et

2005-2010

AC C

Cannon 2012

Japan

M AN U

OLR Beppu et al, 2015

No. of patients

SC

Study (author/year)

RI PT

Table 1 Baseline characteristics of studies included in the meta-analysis

liver/Simul

71 (53.4)

>5 (n= 21) >5 (n= 27) mean=2.6

liver/Simul mean= 3.0

No. of tumor

median= 1(1-4) median= 1(1-5)

ACCEPTED MANUSCRIPT

LLR Italy

2005-2012 R

Multi-center

mean= 3.0

55 (24)

mean= 3.0

60±11

6 (24)

mean= 2.9

50

62±12

10 (20)

mean= 3.4

LLR

89

67±11

7(8)

mean= 2.8

OLR

89

65±9

5 (6)

mean= 2.9

LLR

18

68±6

2 (11.1)

mean= 2.8

OLR

18

64±3

3 (16.7)

mean= 4.25

OLR et al, 2016

Italy

LLR 2004-2015 R

Single-center

liver

France

Untereiner et al, France 2015

1997-2013 R

2012-2015

R

Multi-center

Multi-center

25

liver

liver/Simul

TE D

et al,

M AN U

OLR Tranchart 2015

225

SC

Ratti

75 (70–87) 75 (70–86)

47 (21)

liver/Simul

225

RI PT

Martı´nez-Cecilia et al, 2017

Multiple (n= 92) Multiple (n= 92) median= 2 (1–6) median= 2 (1–7) median= 1.4 median= 1.5 median= 1 median= 1

AC C

EP

69 mean= 2.9 NA 0 (0) 2008-2016 (65-75) Zeng et al, 2016 China R Single-center liver 68 OLR 79 0 (0) mean= 3.3 NA (65-76) R: retrospective design PSM: matching based on propensity score; NA: not available；LLR: laparoscopic liver resection; OLR: open liver resection; liver/simul: a portion of patients in the study underwent liver resection and the other patients underwent synchronous hepatectomy and colectomy; simul: all the patients in the study underwent synchronous hepatectomy and colectomy; MH: major hepatectomy LLR

79

ACCEPTED MANUSCRIPT

Allard et al

Outcome of Interest

*

*

*

Beppu et al

*

*

*

*

Cannon et al

*

*

*

*

Cipriani et al

*

*

*

de’Angelis et al

*

*

*

Martı´nez-Cecilia et al

*

*

*

Ratti

*

*

*

*

*

*

*

*

*

et al

Untereiner et al

Zeng et al

Total Score

…

*

8

*

…

*

7

*

*

*

*

8

*

**

*

*

*

9

*

*

*

*

*

8

*

**

*

*

*

9

*

*

*

*

*

8

*

*

*

*

…

7

*

*

…

*

*

…

6

*

*

*

*

*

*

8

TE D

EP

Tranchart

Outcome Length of Adequacy Follow-up of Follow-up

*

*

AC C

et al

** *

M AN U

*

Comparability Assessment of Outcome

SC

Exposed Cohort

Study

Selection Nonexposed Ascertainment Cohort of Exposure

RI PT

Table 2 Methodological Quality Assessment of Included Studies by Newcastle–Ottawa Scales

ACCEPTED MANUSCRIPT

Table 3 Subgroup analysis

WMD/OR

5 5 3 7 3 7 3 7 2 8

0.53 0.56 0.52 0.59 0.87 0.49 0.36 0.64 0.49 0.58

95 % CI 0.38 0.43 0.39 0.44 0.55 0.32 0.25 0.44 0.34 0.37

SC

RI PT

No. of studies

M AN U

Comparison Overall morbidity Studies with single-center Studies with multi-center Studies with sample (>300) Studies with sample (<300) Studies with simul Studies with liver/simul Studies with major hepatectomy (>40％) Studies with major hepatectomy (<40％) Studies with only elderly patients Studies with mixed-age patients

0.75 0.73 0.69 0.79 1.38 0.73 0.52 0.95 0.70 0.92

P value

I2(%)

0.0003 <0.0001 <0.00001 <0.00001 0.57 0.00006 <0.00001 0.03 0.0001 0.02

0 76 84 26 0 62 0 29 0 64

CI: confidence interval; WMD: weighted mean difference; OR: odds ratio; I2: degree of statistical heterogeneity (0–25 % moderate, 26–50 % average,

TE D

51–100 % high statistical heterogeneity); PSM: matching based on propensity score; liver/simul: a portion of patients in the study underwent liver resection and the other patients underwent synchronous hepatectomy and colectomy; simul: all the patients in the study underwent synchronous

AC C

young; P values of 0.01 include all data of <0.01.

EP

hepatectomy and colectomy; MH: major hepatectomy; mixed-age: a portion of patients in the study were elderly and the other were middle-aged or

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT FIG. 1 illustrates the study screening and selection process.

FIG. 2 Forest plots depicting operative time, blood loss and blood transfusion in the included studies. ORs and WMDs are shown with 95 % CIs.

included studies. ORs and WMDs are shown with 95 % CIs.

RI PT

FIG. 3 Forest plots depicting overall morbidity, 90-day mortality and hospitalization time in the

M AN U

studies. ORs and WMDs are shown with 95 % CIs.

SC

FIG.4 Forest plots depicting surgical margins R0, recurrence and 3-year DFS in the included

FIG. 5 Forest plots depicting 3-year DFS, 3-year OS and 3-year OS reported in the included studies. ORs are shown with 95 % CIs

AC C

EP

TE D

FIG. 6 Funnel plots depicting the publication bias in the included studies

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Highlights To compare the short-term and oncological outcomes of patients with colorectal liver metastases (CRLM) who underwent laparoscopic liver resection (LLR) versus open liver resection (OLR) LLR is a beneficial alternative to OLR in selected patients.