3.3. Primary endpoints: 5-year OS and DFS.
Four studies reported the 5-year OS [9–12]. Three studies directly provided HR values for OS [9, 10, 12], and another study presented the results as a step-like survival curve [11]. We extracted the statistics according to the method provided by Tierney et al. using Engauge Digitizer software [21]. Four studies, with a total of 2189 patients, were pooled for analysis, and the random effects model was used because high heterogeneity was observed (I²=50%, P = 0.11). The results showed no significant difference between the TME with LLND group and TME alone group (pooled HR 0.93, 95% CI 0.71–1.22, P = 0.62). Subgroup analysis showed no significant difference between the two arms regardless of the use of preoperative nCRT (HR = 1.41, 95% CI 0.56–3.55, P = 0.47 vs HR = 0.90, 95% CI 0.68–1.20, P = 0.42, respectively). The details are shown in Fig. 4a.
Data on 5-year DFS were reported in six studies. Three studies provided HR values of 5-year DFS directly [2, 9, 12], and the results of another two studies were extracted from the step-like survival curve provided by the authors [11, 24], We excluded the study of Dev, K because the results were 2-year disease-free survival [15]. Ultimately, five studies that included 1552 patients were pooled into the analysis [2, 9, 11, 12, 24]. The random effects model was used because of the high between-study heterogeneity (I²=50%, P = 0.08). The results indicated no significant difference between the TME with LLND group and TME alone group (pooled HR 0.99, 95% CI 0.74–1.34, P = 0.96). The subgroup analysis showed no significant difference between the two arms regardless of the use of nCRT (HR = 0.71, 95% CI 0.40–1.25, P = 0.23 vs HR = 1.08, 95% CI 0.75–1.55, P = 0.69, respectively). The details are shown in Fig. 4b.
3.4. Secondary endpoints: total, local, lateral, and distant recurrence, operation time, intraoperative blood loss, postoperative complications, peri-operative mortality, sexual and urinary dysfunction.
Four studies with a total of 1107 patients were eligible for the analysis of total recurrence [2, 9, 11, 24]. Two of these studies were RCTs [9, 11], and the other two studies were retrospective [2, 24]. The fixed effects model was used to pool the statistics (I²=0%, P = 0.67). No significant difference in total recurrence was found between the two groups (pooled RR 0.98, 95% CI 0.81–1.18, P = 0.83). The subgroup analysis showed no significant difference between the two arms with or without nCRT (RR = 1.46, 95% CI 0.76–2.81, P = 0.25 vs RR = 0.94, 95% CI 0.77–1.14, P = 0.53, respectively). The details are shown in Fig. 5a.
Seven studies with a total of 3220 patients were included in the study of local recurrence [2, 6, 9, 11, 12, 16, 24]. The fix effects model was used to pool the statistics (I²=24%, P = 0.24). The results indicated that local recurrence rate was significantly lower in the TME with LLND group than the TME alone group (pooled RR 0.71, 95% CI 0.56–0.89, P = 0.003). The subgroup analysis found that the LLND group had a significantly lower incidence of local recurrence than the TME alone group when preoperative nCRT was not performed (RR 0.71, 95% CI 0.56–0.89, P = 0.004). However, the difference was not significant once preoperative nCRT was introduced (RR 0.70, 95% CI 0.32–1.51, P = 0.36). The details are shown in Fig. 5b.
Data on lateral recurrences rate were extracted from three studies of 2369 patients [6, 9, 16]. The random effects model was used because of the high heterogeneity (I²=66%, P = 0.05). The results demonstrated no significant difference in lateral recurrence rate between the two groups (pooled RR 0.49, 95% CI 0.18–1.28, P = 0.14). The subgroup analysis indicated no significant difference between the two arms regardless the introduction of nCRT (RR = 0.72, 95% CI 0.27–1.97, P = 0.53 vs RR = 039, 95% CI 0.08–1.89, P = 0.24, respectively). The details are shown in Fig. 5c.
Distant recurrence was reported in five studies that investigated 1819 patients [2, 11, 12, 16, 24]. There was no significant difference between the two groups (pooled RR 0.95, 95% CI 0.68–1.34, P = 0.78). The random effects model was used because of moderate between-study heterogeneity (I²=43%, P = 0.12). The subgroup analysis revealed no significant difference between the two arms regardless of the use of preoperative nCRT (RR = 0.74, 95% CI 0.41–1.33, P = 0.32 vs RR = 1.14, 95% CI 0.89–1.47, P = 0.29, respectively). The details are shown in Fig. 5d.
Four studies were included in the meta-analysis that assessed the length of operation in 1195 patients [2, 11, 15, 25]. The results demonstrated a significant difference that favored the non-LLND group (WMD 90.73 min, 95% CI 75.35-118.72, P < 0.001) with significant between-study heterogeneity (I²=96%, P < 0.001), and the random effects model was used. The subgroup analysis revealed that the TME with LLND group needed a longer operation time than the TME alone group regardless of the use of preoperative nCRT (WMD = 79.85, 95% CI 74.81–84.88, P < 0.001 vs WMD = 110.74, 95% CI 93.55-127.95, P < 0.001, respectively). The details are shown in Fig. 6a.
Four studies were included in the meta-analysis to assess intraoperative blood loss in 1195 patients [2, 11, 15, 25]. The random effects model was used because of the high between-study heterogeneity (I²=99%, P < 0.001). The results indicated that the TME alone group showed significantly lower intraoperative blood loss than the TME with LLND group (WMD 303.20 mL, 95% CI 156.82–449.58, P < 0.001). The subgroup analysis found that the TME alone group had significantly lower intraoperative blood loss than the TME with LLND group when nCRT was not performed, but the difference was not significant when nCRT was introduced (WMD = 434.84, 95% CI 34.39–835.30, P = 0.03 vs WMD = 256.86, 95% CI -211.92-725.64, P = 0.28, respectively). The details are shown in Fig. 6b.
Three studies assessed 992 patients and reported postoperative complications [2, 11, 25]. A fixed effects model was used because of the low between-study heterogeneity (I²=0%, P = 0.77). The TME with LLND group was associated with a higher rate of postoperative complications than the TME alone group (pooled RR = 1.35, 95% CI 1.05–1.74, P = 0.02). The subgroup analysis demonstrated more postoperative complications in the TME with LLND group than the TME alone group when nCRT was not undertaken, but the difference was not significant when neoadjuvant therapy was introduced (RR = 1.39, 95% CI 1.05–1.83, P = 0.02 vs RR = 0.1.13, 95% CI 0.65–1.96, P = 0.66, respectively). The details are shown in Fig. 7a.
Peri-operative mortality was reported in three studies that investigated 992 patients [2, 11, 25]. The data extracted from one of the studies was not suitable for meta-analysis because no events occurred in either group [11]. Ultimately, two studies including 947 patients without neoadjuvant therapy were pooled into the analysis [2, 25]. A fixed effects model was used because of the low between-study heterogeneity (I²=0%, P = 0.49). The results indicated no significant difference between the two groups (pooled RR = 1.52, 95% CI 0.18–12.65, P = 0.70). The details are shown in Fig. 7b.
Two RCTs studies assessed 200 patients and reported sexual dysfunction [11, 26]. The random effects model was used because of high between-study heterogeneity (I²=55%, P = 0.13). The results indicated no significant difference between the two groups (pooled RR 1.41, 95% CI 0.87–2.31, P = 0.17). The subgroup analysis demonstrated that the sexual dysfunction rate was lower in the TME alone group when nCRT was introduced, but the difference was not significant when preoperative nCRT was not performed (RR = 2.03, 95% CI 1.00-3.95, P = 0.04 vs RR = 1.19, 95% CI 0.95–1.49, P = 0.13, respectively). The details are shown in Fig. 7c.
Two RCTs studies assessed 746 patients and reported urinary dysfunction [11, 27]. The random effects model was used because of the high between-study heterogeneity (I²=80%, P = 0.03). The results demonstrated no significant difference between the two groups (pooled RR 1.44, 95% CI 0.63–3.28, P = 0.38). The subgroup analysis demonstrated that the urinary dysfunction rate was lower in the TME alone group when nCRT was introduced, but the difference was not significant when nCRT was not performed (RR = 2.39, 95% CI 1.14–5.04, P = 0.02 vs RR = 1.02, 95% CI 0.90–1.16, P = 0.74, respectively). The details are shown in Fig. 7d.