Dissection of the lymph nodes adjacent to the stomach and celiac axis (D2 lymphadenectomy) has been an essential component of the surgical treatment of resectable advanced gastric cancer (AGC) in East Asia for some time, and is now increasingly accepted as the standard of care in Western countries [1]. The extent and quality of lymph node dissection is critical: in laparoscopic-assisted distal gastrectomy (LADG) for AGC, this can be measured by the number of harvested lymph nodes (HLNs) and the incidence of non-compliance (the identification of nodal tissue at a node station that should have been resected) [2–4]. The number of HLNs is frequently used as an outcome measure in studies of LADG, but it remains unclear how the number of HLNs yielded by D2 lymphadenectomy during LADG compares with the number yielded during open distal gastrectomy (ODG).

Several non-randomized controlled trials (NRCTs) have been undertaken to compare D2 lymphadenectomy during LADG with D2 lymphadenectomy during ODG for the treatment of AGC [5–12]. We undertook a systematic review and meta-analysis of data from these NRCTs. As well as the number of HLNs, we also systematically and objectively compared the body mass index (BMI), tumor size, serosa invasion status, and positive lymph node rate of patients who underwent LADG with those who underwent ODG. The latter four parameters are considered to have contributed to selection bias in these NRCTs. We proposed null hypotheses that the number of HLNs would not be significantly different in patients undergoing laparoscopic or open surgery, and that the absence of significant differences in the other parameters would indicate that there was no selection bias in the recruitment to the NRCTs. If there were significant differences, it would suggest that it could not be convincingly established that the quality of D2 lymphadenectomy undertaken during LADG was equivalent to that undertaken during ODG in patients with AGC.

Methods

Search strategy

Studies were identified by searches of PubMed, EMBASE, Web of Science, and the Cochrane Library for relevant articles published between January 1995 and June 2013. The search strategy was as follows:

PubMed

(“Stomach Neoplasms”[Mesh] AND “surgery”[Subheading]) AND “laparoscopy”[mesh] AND ((“1995/01/01”[PDAT]: “2013/06/30”[PDAT]) AND “humans”[MeSH Terms] AND English(lang)

Embase

‘Stomach’/exp and ‘neoplasm’/exp and ‘laparoscopic surgery’/exp and [humans]/lim and [english]/lim and [embase]/lim and [1-1-1995]/sd not [30-6-2013]/sd

Web of Science

(TS = (gastric cancer AND surgery AND laparoscopy)) AND Language = (English)

Databases = SCI-EXPANDED, CPCI-S, CCR-EXPANDED, IC Timespan = 1995-2013

Cochrane Library

#1 [Stomach Neoplasms] explode all trees and with qualifiers: [Surgery—SU]

#2 [Laparoscopy] explode all trees

#3 #1 and #2 (from 1995 to 2013)

Inclusion and exclusion criteria

We included clinical trials that: (1) made a comparison between LADG with D2 lymphadenectomy and ODG with D2 lymphadenectomy for the treatment of AGC from 1995 to 2013; (2) recorded the majority of the following: number of HLNs, BMI, tumor size, serosa invasion status, and the positive lymph node rate; (3) were written in English; and (4) involved a previously unreported patient group. We excluded trials that: (1) did not use ODG as a concurrent control; (2) provided insufficient data or were duplicate publications.

Definitions

We included studies in which D2 lymphadenectomy was performed according to the lymph node classification of the Japanese Gastric Cancer Association (guideline version 3) [13], which state that lymph node numbers 1, 3, 4sb, 4d, 5, 6, 7, 8a, 9, 11p, and 12a should be dissected. Advanced gastric cancer was defined as invading the muscularis propria of the stomach wall or beyond [14].

Quality of literature

For the purposes of this review, we used a modified scale method to assess the quality of literature according to a previously established scoring system, the methodological index for non-randomized studies (MINORS) [15]. The modifications included: (1) six of the 12 items were not considered: four were not related to the validity of the data (statement of the study aim, evaluation of endpoints, and adequacy of statistical methods) and two were better suited to assess the validity of long-term outcomes (adequate follow-up period and follow-up loss <5 %); (2) the item “prospective calculation of sample size” was modified to assess the proficiency of the surgical teams and the power given by the actual number of LADG cases; and (3) the item “important data being presented” was included in the modified MINORS scale (Table 1).

Table 1 Modified MINORS scale used for quality assessment of non-randomized controlled trials
Full size table

Data extraction

Two researchers independently extracted data from each study. The extracted information comprised: study features, data needed for quality assessment, and the patients’ clinical pathological parameters (including HLN, BMI, tumor size, tumor stage, and number of positive lymph nodes). If the study provided medians and ranges instead of means and standard deviations (SDs), we estimated the means and SDs using the technique described by Hozo and colleagues (Table 2) [16].

Table 2 Estimating formulae for an unknown distribution
Full size table

Statistical analysis

Meta-analysis was performed using weighted mean differences (WMDs) for continuous variables and odds ratios (ORs) or risk differences (RDs) for dichotomous variables. The latter were used if there were no events in either the intervention or control arms of the study [17]. Between-study heterogeneity was assessed by performing χ 2 tests (assessing the P value), and by calculating the I 2 statistic. If the P value was less than 0.05 and I 2 exceeded 50 %, we considered heterogeneity to be substantial. In these circumstances, random-effect models were used instead of fixed-effect models. Ninety-five percent confidence intervals (CIs) are reported. All reported P values are two sided, and values <0.05 are considered statistically significant. Statistical analyses were performed using Stata Version 11 (Stata Corp, College Station, Tex., USA).

Results

Studies selected

The search strategy retrieved 1,359 reports in English. After the titles and abstracts had been reviewed, 29 reports were included for full-text review (six randomized controlled trials and 23 NRCTs). Of these, 19 were excluded because tumor stage or control data were unavailable. Ten reports were finally selected for quality assessment. A flow chart of the search strategies is illustrated in Fig. 1.

Fig. 1
figure 1

Article selection flow chart. RCT randomized controlled trial, NRCT non-randomized controlled trial, MINORS methodological index for nonrandomized studies

Full size image

Quality assessment

The quality assessment of the NRCTs is shown in Table 3. The studies by Huscher et al. and Cui et al. were excluded because their score was <10 points.

Table 3 Modified MINORS score of all eligible non-randomized comparative studies
Full size table

Outcomes

Number of HLNs

Two of the eight studies reported that there were statistically significant differences between the number of HLNs in the LADG and ODG groups, but in the pooled data, there was no significant difference between the groups (WMD: −0.98, 95 %CI −3.21 to 1.26; Fig. 2A)

Fig. 2
figure 2figure 2figure 2

A Meta-analysis of the number of harvested lymph nodes in LADG versus ODG. B Meta-analysis of body mass index in LADG versus ODG. C Meta-analysis of tumor size in LADG versus ODG. D Meta-analysis of serosa invasion rate in LADG versus ODG. E Meta-analysis of rate of positive lymph nodes in LADG versus ODG

Full size image

BMI

Only one of three studies found that BMI was significantly lower in the LADG group than the ODG group. In the pooled data, there was no statistically significant difference (WMD: −1.20, 95 % CI −2.64 to 0.24; Fig. 2B).

Tumor size

Two of five studies demonstrated that the tumor size was statistically different between LADG and ODG, but there was no significant difference in the pooled data (WMD: −0.30, 95 % CI −0.65 to 0.05; Fig. 2C).

Serosa invasion status

In the pooled data, there was no significant difference in serosa invasion rate between the groups (RD: 0.04, 95 % CI −0.03 to 0.11; Fig. 2D).

Rate of positive lymph nodes

In the pooled data, there was no significant difference in the rate of positive lymph nodes between the groups (OR: 0.66, 95 % CI −0.44 to 1.01; Fig. 2E).

Discussion

Laparoscopic resection is a technique that offers important advantages (less blood loss, reduced postoperative pain, accelerated recovery, early return to normal bowel function, and reduced hospital stay) over open surgical procedures for patients with gastric cancer [20]. At the same time, some prospective randomized studies have shown that the survival rates of patients with early stage cancers in the middle or lower parts of the stomach undergoing either LADG or ODG were broadly comparable [21, 22]. This strongly suggests that LADG is an effective treatment for early gastric cancer.

The efficacy of LADG for the treatment of AGC is less clear. Because of the complexity of the laparoscopic technique, surgeons must focus on the quality of D2 lymphadenectomy during LADG, which is a fundamental aspect of oncological safety in the era of “D2 surgery” for AGC [2, 4]. In most clinical studies, the number of HLNs is widely accepted as the index of “quality” [2, 6, 18, 19, 21].

Our literature review identified 10 NRCTs that compared the efficacy of LADG with ODG for the treatment of AGC. We identified no randomized controlled studies. The modified MINORS scale method was used to assess the quality of these reports and eight of which scored no less than 10 were finally included in this study [15]. We focused on not only the number of HLNs but also the other clinicopathological parameters including BMI, tumor size, serosa invasion status, and positive lymph node rate. The latter four items were used to assess whether there had been selection bias in enrollment for the studies analyzed. Patients’ age and sex were not included in our meta-analysis as they were considered unlikely to influence the difficulty of lymph node dissection.

Our meta-analysis revealed that there were no significant differences in the number of HLNs, BMI, tumor size, and the rates of serosa invasion and positive lymph nodes between the groups. This suggests that the extent of obesity and tumor stage of enrolled patients, as well as the number of HLNs, was comparable between the groups and thus a conclusion can be drawn that the quality of D2 lymphadenectomy at LADG is equal to that at ODG for patients with AGC.

This is the first meta-analysis study to have focused on the number of HLNs from this perspective, which we considered to be not only an important predictor of oncological outcome after LADG, but also allowed a re-assessment of the quality of the non-randomized studies themselves. The findings of high-quality NRCTs might be as reliable as randomized controlled trials, particularly when pooled data are compared to evaluate the effectiveness of surgical procedures [23]. Nonetheless, our study has some limitations: no original data were obtained and no conclusions could be drawn about publication bias, as only eight studies were included. A high-quality randomized study is still necessary to allow firm conclusions to be drawn.