Study selection
Duplicated records were deleted. We deleted 1843 studies after reading the titles and abstracts carefully. Deleted studies were due to not rectal cancer(n=527), confusing group(n=851) and insufficient data(n=455). After we read the remaining studies carefully, 14 English studies with 3932 patients were included [6, 11-23].There were 3 patients in the WS group and 1 patient in the surgical group with Stage IV. There are 700 patients in WS group and 3232 patients in surgical group. The clinical stages of included patients areⅠto IV stage. LE groups did not included patient with stage 0 and/or Stage IV. Surgical group divided into radical surgical group(n=3140) and LE group (local excision group, n=92) (Table 1). Two groups had similar patient baseline with no significant difference. There were 6 Eastern studies and 8 Western studies in the meta-analysis. The Western research includes European, American, and Latin American research, while the Eastern research mainly includes Asian research. Martens, Yeom, Lin and Lai reported local resection for rectal cancer with cCR/near-cCR response after neoadjuvant chemoradiotherapy. Rectal cancer patients which performed radical surgery and local excision were put together for analysis in Lin’s research, the patient outcome of two surgery subgroups was not described separately. So there were only 92 patients data could be collected. Martens described near-cCR in his study, there were patients with 39 near-cCR and 61 patients with cCR in both groups. There is no possibility in patient profiles among WS, RS and LE groups.There was no statistical difference in the baseline data of the included patients for each article.
Most of the studies have used the above five criteria to determine whether patients have achieved cCR or near-cCR[16-23]. Araujo used endoscopy, DRE and MRI to confirm cCR[11]. Ayloor used endoscopy, DRE, CT and ultrasound to confirm cCR[12]. Dalton used endoscopy and MRI to confirm cCR[13]. Habr used endoscopy, DRE and CT to confirm cCR[6]. Lai used endoscopy and MRI to confirm cCR[14]. Lee used DRE and MRI to confirm cCR[15]. The detail was shown in table 1.
Quality assessment
We evaluated the quality of the included studies using NOS assessment scale (Newcastle Ottawa Quality Assessment Scale). The qualities of included studies were divided into three levels: low level (1<scores≤3), moderate level (4< scores≤6) and high level (7≤scores≤9). We included 14 studies in our study (5 RCNT and 9 PNCT). No relevant RCTs was been found during the database search. Seven studies were of moderate quality and seven studies of high quality. The quality scores and study type were shown in table 1.
WS group versus Surgery group
Local recurrence, Distant metastasis and Cancer related death
12 studies reported clinical data on local recurrence, WS group had higher recurrence rate than surgery group in the fixed-effects model (OR:3.55, 95% confidence interval [CI]:2.35 to 5.36, P<0.001, chi2=9.9, P=0.54, I2=0%, Fig 2a). Both Western (OR:3.14, 95% CI:1.31 to 7.52, P=0.01, chi2=8.14, P=0.15, I2=39%, Fig 2a) and Eastern studies (OR:3.74, 95% CI:2.39 to 5.85, P<0.01, chi2=1.39, P=0.92, I2=0%, Fig 2a) had the same results in local recurrence. Distant metastasis (OR:0.94, 95% CI:0.68 to 1.29, P=0.69, chi2=14.76, P=0.19, I2=25%, Fig 2b) and cancer related death (OR:0.77, 95% CI:0.32 to 1.84, P=0.56, chi2=5.54, P=0.24, I2=28%, Fig 2c)were similar between two groups in the fixed-effects model with little heterogeneity.
WS group versus Surgery group
2-year DFS, 2-year OS, 5-year DFS and 5-year OS
10 studies reported 2-year DFS and WS group had better 2-year DFS than surgery group in the fixed-effects model with little heterogeneity (OR:0.74, 95% CI:0.56 to 0.96, P=0.03, chi2=9.01, P=0.44, I2=0%, Fig 3a). 9 studies reported 2-year OS and WS group had better 2-year OS than surgery group in the fixed-effects model with little heterogeneity (OR:0.38, 95% CI:0.28 to 0.52, P<0.001, chi2=13.89, P=0.05, I2=50%, Fig 3b). Eastern research rather than Western research indicated 2-year DFS and 2-year OS were better in WS group than in surgical group. 5-year DFS (OR:0.46, 95% CI:0.19 to 1.12, P=0.09, chi2=20.55, P=0.001, I2=76%, Fig 3c) and 5-year OS (OR:0.79, 95% CI:0.29 to 2.14, P=0.65, chi2=11.19, P=0.05, I2=55%, Fig 3d)were similar in both groups in the random-effects model with huge heterogeneity.
Subgroup analysis
WS group versus Radical surgery group
Local recurrence and Distant metastasis
11 studies reported clinical data on local recurrence, WS group had higher recurrence rate than radical surgery group in the fixed-effects model (OR:4.71, 95% CI:3.07 to 7.21, P<0.001, chi2=4.1, P=0.94, I2=0%, Fig 4a). Both Western (OR:7.47, 95% CI:2.17 to 25.64, P=0.001, chi2=2.13, P=0.71, I2=0%, Fig 4a) and Eastern studies (OR:4.18, 95% CI:2.65 to 6.58, P<0.001, chi2=1.26, P=0.94, I2=0%, Fig 4a) had the same results in local recurrence. WS group had similar distant metastasis rate with radical surgery group in the fixed-effects model with high heterogeneity (OR:0.97, 95% CI:0.70 to 1.35, P=0.87, chi2=14.57, P=0.15, I2=31%, Fig 4b).
WS group versus local excision group
Local recurrence and Distant metastasis
Similar distant metastasis rate was in two groups in the fixed-effects model with little heterogeneity(OR:1.10, 95% CI:0.46 to 2.63, P=0.84, chi2=2.5, P=0.29, I2=20%, Fig 4c). 3 studies reported clinical data on local recurrence, local recurrence was similar in two groups in the fixed-effects model (OR:0.78, 95% CI:0.26 to 2.36, P=0.66, chi2=3.05, P=0.22, I2=34%, Fig 4d).
WS group versus Radical surgery group
2-year DFS and 5-year DFS
10 studies reported 2-year DFS and WS group had better 2-year DFS than surgery group in the fixed-effects model with little heterogeneity (OR:0.73, 95% CI:0.56 to 0.95, P=0.02, chi2=9.48, P=0.39, I2=5%, Fig 5a). Eastern research rather than Weatern research had the same result (OR:0.70, 95% CI:0.51 to 0.96, P=0.03, chi2=0.75, P=0.69, I2=0%, Fig 5a). 5-year DFS was similar in both groups in the random-effects model with huge heterogeneity (OR:0.44, 95% CI:0.18 to 1.12, P=0.09, chi2=21.86, P<0.001, I2=77%, Fig 5b).
WS group versus Local excision group
3-year DFS
2 studies indicated WS group and local excision group had the similar 3-year DFS in the fixed-effects model (OR:0.35, 95% CI:0.09 to 1.29, P=0.11, chi2=0.33, P=0.56, I2=0%, Fig 5c).
WS group versus Radical surgery group(cCR)
Local recurrence, Distant metastasis and Cancer related death
Due to only one research about near-cCR(Martens), the result of local recurrence and distant metastasis were the same to fig 4a and fig 4b (WS group versus Radical surgery group). The result of cancer related death was the same to fig 2c (WS group versus Radical surgery group).
WS group versus Radical surgery group(cCR)
2-year DFS, 2-year OS, 5-year DFS and 5-year OS
Due to no available information of the only one near-cCR research (Martens), the result of 2-year DFS, 2-year OS, 5-year DFS and 5-year OS were the same to fig 3a, fig 3b, fig 3c and fig 3d(WS group versus Surgery group).
WS group versus local excision group(cCR)
Local recurrence and Distant metastasis
2 studies indicated WS group and local excision group had the similar local recurrence (OR:1.46, 95% CI:0.49 to 4.35, P=0.5, chi2=1.69, P=0.19, I2=41%, Fig 5d) and distant metastasis (OR:1.24, 95% CI:0.33 to 4.69, P=0.75, chi2=0.94, P=0.33, I2=0%, Fig 5e)in the fixed-effects model.
There was one research with cCR which perform local excision, it can not be studied by meta-analysis. There was one research with near-cCR which perform local excision, so it can not be studied by meta-analysis too. Due to the lack of related research about the near-cCR status, there is no large amount of data for meta-analysis.
Publication bias
In order to determine whether the article had publication bias, we used Revman5.0 software to test the index of distant metastasis rate of the included literature and obtained the funnel plot. The points were evenly distributed in the funnel plot and it indicated no publication bias in the meta-analysis.