Pooled analysis of the oncological outcomes in robotic gastrectomy versus laparoscopic gastrectomy for gastric cancer.

AIM: Robotic gastrectomy (RG) is more and more widely used in the treatment of gastric cancer. However, the long-term oncological outcomes of RG have not been well evaluated. The aim of this study was to evaluate the long-term oncological outcomes of RG and laparoscopic gastrectomy (LG) in the treatment of gastric cancer.
MATERIALS AND METHODS: PubMed, China National Knowledge Infrastructure, Cochrane Library and EMBASE electronic databases were searched until August 2019. Eligible studies were analysed for comparison of oncological outcomes between RG and LG in patients with gastric cancer.
RESULTS: Eleven retrospective comparative studies, which included 1347 (32.52%) patients in the RG group and 2795 (67.48%) patients in the LG group, were selected for the analysis. Meta-analysis of the 11 included studies showed that there was no statistically significant difference in the OS between the RG and LG groups (hazard ratios [HRs] = 0.97, 95% confidence intervals [CIs] = 0.80-1.19, P = 0.80). Six studies evaluated disease-free survival (DFS), and pooled analysis showed that there was no statistically significant difference in DFS between RG group and LG group (HR = 0.94, 95% CIs = 0.72-1.23, P = 0.65). According to the odds ratio (OR) analysis, there was no significant difference in 3-year OS, 5-year OS, 3-year DFS and 5-year DFS between the RG and LG groups. Nine articles reported the recurrence rate, and the meta-analysis showed that there was no statistically significant difference between the RG and LG groups (OR = 0.88, 95% CIs = 0.69-1.12, P = 0.31).
CONCLUSIONS: This meta-analysis indicated that the long-term oncological outcomes in the RG group were similar to that in the LG group.

INTRODUCTION

Stomach cancer is the fifth-most common malignant tumour and the third-largest cause of cancer-related death worldwide.[1] In general, half of the total global incidences occurs in Eastern Asia, especially in China. It was estimated that there were about 679,100 newly diagnosed stomach cancer cases and 498,000 deaths due to stomach cancer in 2015 in China.[2] Now, there are three common methods of surgical treatment for gastric cancer: open gastrectomy, laparoscopic gastrectomy (LG) and robotic gastrectomy (RG). In 1994, successful laparoscopy-assisted distal gastrectomy with lymph node dissection for early gastric cancer (EGC) was introduced by Kitano et al.[3] LG, as a feasible surgical method, has become an effective method for the treatment of gastric cancer. Despite the advantages of LG over open surgery in gastric cancer[45] there are some limitations, for example, the natural tremor of the hand, lack of flexibility, the two-dimensional surgical field of vision and decreased sense of touch.[678] In addition, LG has a longer learning curve and causes more fatigue for surgeons.[9]

Now, robotic surgical systems overcome LG limitations by providing a three-dimensional magnified view, more comfortable surgeon position, ergonomic comfort and elimination of physiologic tremor.[610] Since the first robot-assisted distal gastrectomy was performed by Giulianotti et al. in 2003,[11] more and more studies have shown that RG is a safe and feasible method for the treatment of gastric cancer.[121314] However, most of these studies were retrospective and focused on the results of short-term surgical procedures.[151617] To date, most systematic reviews and meta-analyses have focused on comparing the efficacy and feasibility of RG and LG in gastric cancer.[181920] Therefore, the long-term oncological results of RG need to be further evaluated.[21]

RG has been performed for nearly 20 years, but there is no better evaluation of long-term gastric cancer treatment. In 2017, Pan et al. reported that there was no significant difference in the overall survival (OS) and disease-free survival (DFS) between RG and LG groups, which included five studies and involved 1614 patients.[22] Recently, some new literature have been published. Hence, we performed this updated meta-analysis to study the long-term OS, DFS and post-operative recurrence rate in patients with gastric cancer. We hope to be able to produce an updated and revised summary of the current evidence for surgeons, and to make surgical practice more in line with current evidence.

MATERIALS AND METHODS

Literature search strategy

Search was applied to the following four electronic databases: PubMed, EMBASE, Cochrane Library databases and China National Knowledge Infrastructure (the last search was updated on 10 August 2019). The following terms were used: 'Robotic Surgical Procedures' [MeSH Terms] OR 'Robotic' [MeSH Terms] AND 'laparoscopy' [MeSH Terms]) AND 'gastrectomy' [MeSH Terms] and 'Stomach Neoplasms'. The search was performed in studies conducted on human subjects, with restriction on English and Chinese languages. We did not consider abstracts or unpublished reports. The reference lists of reviews and retrieved articles were hand-searched at the same time. This meta-analysis was carried out in accordance with the Preferred Reporting of Systematic Reviews and Meta-Analyses.[23] The Methodological Index for Non-randomized Studies (MINORS) scoring system was used to assess the methodological quality of the identified studies.[24] The score for the items varies from 0 to 2, where 0 means no report, 1 means insufficient information being reported and 2 means the report and sufficient information being provided. The highest score of the MINORS is 24. Each item was evaluated by two authors (Yang Ping and Lin Xiu-Feng) and a consensus was reached in case of disagreement. The review was registered in PROSPERO in 2019 (id CRD42019121834).

Inclusion and exclusion criteria

We reviewed abstracts of all citations and retrieved studies. The following criteria were used to include published studies: (a) studies focusing on patients with gastric cancer; (b) comparative studies between RG and LG and (c) studies that reported the outcomes, including OS, DFS or the recurrence rate. OS refers to the patients who were alive from the time of surgery to the date of last follow-up in the study, whereas DFS means patients in the study were alive from the time of surgery to the date of last follow-up without the development of the local or distant disease recurrence or a new gastric tumour. The major reasons for exclusion of studies were (1) case report, reviews and duplicates; (2) the studies in which only non-oncologic outcomes were reported and (3) non-Chinese and English literature.

Data extraction

Data were drawn from each study by two reviewers (Yang Ping and Lin Xiu-Feng) independently according to the pre-specified selection criteria. Any disagreements during screening and quality assessment were resolved by discussion.

Main and secondary outcomes

We considered the following main outcomes of two different surgical procedure groups: OS and DFS. The recurrence rate was recorded as a secondary outcome for the current meta-analysis.

Statistical analysis

The statistical analysis was performed using RevMan5.3 software (London, SW1Y 4QX, UK), which was provided by the Cochrane Collaboration. As for dichotomous variables, the odds ratios (ORs) were calculated, reporting 95% confidence intervals (CIs). Heterogeneity was checked by the Chi-square test. If the results of the trials had heterogeneity, a random-effects model was used for meta-analysis. Otherwise, a fixed-effects model was used. If the survival data were provided with hazard ratios (HRs) and 95% CIs, we directly extracted the data for meta-analysis. In addition, if the survival data were displayed as a Kaplan–Meier survival curve, we used the Engauge Digitizer software which was provided by the UpdateStar GmbH (Kreuzbergstr, Berlin, Germany) to estimate HRs and 95% CIs and rebuild HR and standard deviation (standard error). P < 0.05 was considered statistically significant. The pooled result was expressed with HR or OR with its corresponding 95% CIs.

RESULTS

Study characteristics

There were 512 articles relevant to the search terms [Figure 1]. Through the steps of filtering the title, abstracts and full text, 11 articles were found to conform to our inclusion criteria finally.[2526272829303132333435] The 11 articles (4142 cases) included 1347 (32.52%) patients in the RG group and 2795 (67.48%) patients in the LG group. Included studies were mainly from Northeast Asian countries, including Korea, China, Japan and Italy. Post-operative follow-up varied from 1 month is too short for comparing DFS to 126 months. Characteristics of the studies included in this meta-analysis are presented in Table 1.

Figure 1

Studies identification, inclusion, and exclusion

Table 1

Characteristics of randomised controlled trial studies included in this meta-analysis

Authors (years)	Country	Study period	Number of patients		Mean age (RG/LG) (years)	Duration of follow up (months)	Design
			RG	LG
Pugliese (2010)	Italy	2000.6-2009.10	18	52	71	53 (3-112)	Retrospective
Son (2014)	Korea	2003.5-2010.12	51	58	55.3/58.8	70 (24-112)	RP
Zhou (2014)	China	2010.3-2013.5	120	394	54.7/55.6	RG 15 (3-41),LG19 (3-41)	Retrospective
Lee (2015)	Korea	2003-2010	133	267	53.6/59.2*	75 (36-126)	RP
Nakauchi (2016)	Japan	2009.9-2012.12	84	437	64/68*	RG 40.5 (1.1-75.8),LG 42.2 (1.7-78.9)	Retrospective
Obama (2018)	Korea	2005.7-2009.12	315	525	54.5/59.3	85 (60-114)	Retrospective
Zhang (2018)	China	2011.11-2013.12	27	62	59.7/56.6	52 (18-61)	Retrospective
Li (2018)	China	2013.8-2017.3	125	329	55.4/56.9	28 (3-52)	Retrospective
Gao (2019)	China	2011.1-2014.12	163	339	60.27/59.36	50.5 (36-72)	RP
Xu (2019)	China	2016.6-2018.6	241	231	57/57	19 (7-31)	Retrospective
Ye (2019)	China	2015.1-2016.10	70	101	57/58	25 (6-37)	Retrospective
MINORS score	TNM stage (I/II/III/IV)		Extent of resection (SG/TG)		Lymphadenectomy (D1/D2)	Type of reconstruction (RY/BI/BII)	LN dissection number
16	50/8/9/3		SG		D2	NR	RG (25±4.5), LG (31±8)
18	RG (35/8/8/0), LG (43/10/5/0)		SPT		D2	NR	RG (47.2), LG (42.8)
19	RG (29/36/55/0), LG (115/98/181/0)		RG (94/26), LG (276/118)		D1 or D2	RG (26/1/91), LG (118/7/254)	RG (34.6±10.9), LG (32.7±11.2)
17	RG (101/15/17/0), LG (218/32/17/0)		SG		D2	NR	RG (41.2±13.1), LG (39.9±13.3)
17	RG (61/10/12/1), LG (310/80/42/5)		RG (57/27), LG (301/136)		RG (35/4), LG (232/205)	RG (30/27/27), LG (183/165/89)	RG (44±57.8), LG (38±74.8)
17	RG (277/31/7/0), LG (455/66/4/0)		RG (229/86), LG (422/103)		RG (185/130), LG (316/209)	NR	RG (40.1±15.4), LG (38.6±14.5)
17	RG (24/3/0/0), LG (61/1/0/0)		SG		D1+	NR	RG (30.3±6.2), LG (26.9±6.3)
16	RG (24/51/50/0), LG (37/127/165/0)		SG, TG		D2	NR	RG (29.5±9.6), LG (27.7±8.7)
16	RG (0/57/106/0), LG (0/169/170/0)		RG (101/62), LG (238/111)		RG (56/107), LG (119/220)	NR	RG (30.55±10.13), LG (25.28±9.89)*
16	RG (22/38/181/0), LG (20/39/172/0)		RG (134/107), LG (137/94)		D2	RG (107/0/134), LG (94/0/137)	RY: RG (45±14), LG (39±14)*, BII: RG (36±18), LG (30±16)*
16	RG (16/31/23/0), LG (24/50/27/0)		SG		D2	RG (0/29/41), LG (0/54/47)	RG (22±4), LG (20±4)*

*P<0.05. RG: Robotic gastrectomy, LG: Laparoscopic gastrectomy, LN: Lymph nodes, RP: Retrospective design and prospective data collection, MINORS: Methodological index for non-randomised studies, SG: Subtotal gastrectomy, TG: Total gastrectomy, SPT: Spleen-preserving total gastrectomy, NR: No report, TNM: Tumour, node, metastasis

Quality of the included studies

All the 11 studies were of retrospective design, of which three[262833] were conducted with prospective data collection. The included studies were of a moderate methodological quality, with the MINORS score ranging from 16 to 19 points.

Overall survival outcomes

All the 11 included studies reported OS and provided a high-quality Kaplan–Meier survival curve based on follow-up results. Useful data were generated from these high-quality Kaplan–Meier survival graphs. Meta-analysis of the 11 included studies showed that there was no statistically significant difference in OS between RG and LG groups (HR = 0.97, 95% CIs = 0.80–1.19, P = 0.80) and no significant difference in heterogeneity between the two groups (χ2 = 2.51, P = 0.99, I2 = 0%) [Figure 2].

Figure 2

Overall survival outcomes for robotic gastrectomy versus laparoscopic gastrectomy forest plot

According to the raw data of the literature included, the 3-year survival rate in this meta-analysis was not statistically significant between the RG and LG groups (OR = 0.98, 95% CIs = 0.79–1.22, P = 0.85), and there was no heterogeneity between the two groups (χ2 = 11.11, P = 0.27, I2 = 19%) [Table 2]. Similarly, the 5-year survival results of the pooled analysis showed no statistical significance between the two groups (OR = 0.53, 95% CIs = 0.23–1.22, P = 0.14). The heterogeneity test found that there was statistical significance, so the random mode was chosen (χ2 = 13.19, P = 0.004, I2 = 77%) [Table 2].

Table 2

The results included in this meta-analysis

	Number of studies	OR (95% CI)	P for heterogeneity
The 3-year survival rate	10	0.98 (0.79-1.22)	0.27
The 5-year survival rate	4	0.53 (0.23-1.22)	0.004
The 3-year DFS rate	4	1.17 (0.84-1.64)	0.64
The 5-year DFS rate	2	0.83 (0.49-1.40)	0.59

OR: Odds ratio, CI: Confidence interval, DFS: Disease-free survival

Disease-free survival outcomes

In six studies involving 2515 patients,[262930313233] the Kaplan–Meier survival curve showed DFS outcomes between RG and LG groups. Combined analysis showed that there was no statistical significant difference in DFS between the RG group and LG group (HR = 0.94, 95% CIs = 0.72–1.23, P = 0.65). There was no statistical significant difference in heterogeneity between the two groups (χ2 = 4.22, P = 0.52, I2 = 0%) [Figure 3].

Figure 3

Disease-free survival outcomes for robotic gastrectomy versus laparoscopic gastrectomy forest plot

According to the raw data of the studies included, the 3-year DFS results synthesised between the RG group and the LG group showed no statistical significance (OR = 1.17, 95% CIs = 0.84–1.64, P = 0.35), and there was no heterogeneity between the two groups (χ2 = 1.69, P = 0.64, I2 = 0%) [Table 2]. Similarly, the 5-year DFS results of comprehensive analysis showed no statistical significance (OR = 0.83, 95% CIs = 0.49–1.40, P = 0.48), and no significance of heterogeneity test was found (χ2 = 0.30, P = 0.59, I2 = 0%) [Table 2].

The recurrence rate

Nine studies[252627283031323334] reported the recurrence rate or number of recurrent cases among a total of 3353 long-term follow-up cases in the RG and LG groups. There were 121 cases (11%) in the RG group and 281 cases (12.95%) in the LG group. In terms of gastric cancer recurrence rate, there was no statistically significant difference between the RG group and LG group (OR = 0.88, 95% CIs = 0.69–1.12, P = 0.31), and there was no significant heterogeneity (χ2 = 7.83, P = 0.45, I2 = 0%) [Figure 4].

Figure 4

Forest plot for the recurrence rate between robotic gastrectomy and robotic gastrectomy group

Publication bias and sensitivity analysis

According to the OS rate, the publication bias was evaluated by the funnel plot. All the included studies were within a 95% CI and were symmetrical vertically [Figure 5]. Sensitivity analysis was carried out alternately by random-effects and fixed-effects models. The results showed that, except for the 5-year OS, the other evaluation indicators were consistent with the results of alternate analysis, which indicated that the study was stable, and the analysis results were reliable.

Figure 5

Funnel plot of overall survival

DISCUSSION

With the development of accelerated rehabilitation surgery and surgical injury control, minimally invasive surgery (MIS) has developed rapidly in recent 20 years. As one kind of MIS, robotic surgery system is more and more used by surgeons all over the world. In the treatment of EGC, RG is generally considered an alternative to open gastrectomy now.[36] Radical gastrectomy with lymphadenectomy is now using robotic surgery to treat advanced gastric cancer, and this trend is increasing.[37] Previous studies have focused on the safety and feasibility of RG versus conventional LG in the treatment of gastric cancer.[61112] Although RG has been used for many years in the treatment of gastric cancer, few studies have focused on the long-term follow-up oncological outcomes.[1222] However, for doctors and patients, the long-term oncological outcomes is the most important concern.

In our meta-analysis, 11 retrospective comparative studies were included to analyse OS, DFS and recurrence rates. The mean follow-up time of the included studies was 44.2 months. The results of meta-analysis showed that there was no significant difference in 3-year OS, 5-year OS, OS, 3-year DFS, 5-year DFS, DFS and recurrence rate between RG group and LG group. This result showed that the long-term follow-up oncological outcomes in the RG group were similar to those in the LG group. The results of our meta-analysis were similar to that of previous studies.[1222]

Two studies reported the results of single-centre follow-up.[1238] In a Chinese study,[12] 606 patients with RG for gastric cancer were included, of whom 403 underwent distal gastrectomies, 169 RG total gastrectomies, 15 proximal gastrectomies and 19 remnant gastrectomies. The mean follow-up was 42 months. The cumulative 5-year OS was 69.33%. Another Italian single-centre study,[38] which included 98 patients with RG gastric cancer, reported a 5-year OS rate of 73.3% with a mean follow-up of 46.9 months. This 5-year OS was similar to that of our meta-analysis. It is well known that the OS rate is closely related to the patient's tumour cell differentiation, TNM stage and so on. However, due to the lack of relevant data in the original literature, subgroup meta-analysis cannot be carried out.

In a single-centre study, Jiang et al.[12] reported a 5-year DFS rate of 68.73% for gastric cancer treated with RG. For DFS, our study included almost two and a half times more patients than previous meta-study,[22] but a comprehensive analysis showed no significant difference in DFS between the RG and LG groups. The results suggest that the DFS of gastric cancer treated with RG is similar to that of LG, based on the current evidence. As for tumour recurrence, a retrospective study, including 1198 patients, reported a recurrence rate of 47.2% after LG and a recurrence rate of 50.4% after open gastrectomy.[39] In our meta-analysis, although the post-operative recurrence rates were 11% for RG and 12.95% for LG, the difference was not statistically significant. The results of recurrence rate proved once again that the long-term oncological outcomes of RG and LG in the treatment of gastric cancer were comparable.

Like most meta-analyses, these results should be interpreted with caution. Although the findings of this study have some clinical significance, there were also some limitations. First, the studies included in this meta-analysis were all retrospective in nature, lacking high-quality, randomised controlled, long-term follow-up studies with large samples. Second, the baselines of the two groups of individual studies were inconsistent, such as the average age, which would affect the final analysis results. Third, there is a lack of subgroup analysis, such as follow-up results by tumour stage.

CONCLUSIONS

This meta-analysis shows that RG and LG have similar long-term oncological outcomes in the treatment of gastric cancer. Overall, more randomised clinical trials and a larger cohort of patients with longer follow-up time are still necessary to further demonstrate the value of robotic gastric cancer surgery; in addition, economic evaluation is also needed.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.