Efficacy of minimally invasive tubular approaches for management of the lumbar spinal synovial cysts: a meta-analysis

This article was reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Statement. It was registered at the International Prospective Register of Systematic Reviews (CRD42021288992) [16, 17] (Fig. 1). A comprehensive online search was conducted via Web of Science databases, PubMed and The Cochrane Library on October 17, 2023, using the keywords “spinal synovial cyst”, “spine facet joint cyst”, “paraspinal joint cyst”, “spine degenerative cyst”, “spine ganglion cyst”, “lumbar synovial cysts”. Duplicates and literatures involving synovial cysts of thoracic and cervical levels were excluded. We selected the literature following inclusion and exclusion criteria with no language restrictions. This selection process yielded a total of 41 studies.

Two researchers extracted all baseline data and primary outcomes from each qualified study: patient gender, age, relevant medical history, presenting symptoms, imaging findings, surgical management, and follow-up times. Our primary outcome variables were symptom resolution at last follow-up times and cyst recurrences.

All statistical analyses were performed using Stata 16 (Stata Corp LP) and SPSS statistics 25 (IBM) software. A random-effects model was used to define all pooled outcome measures, and the odds ratio (OR) was estimated with its variance and 95% CI. The prevailing heterogeneity between ORs for the comparable outcomes between different studies was calculated using the I² inconsistency test that depicts the percentage of total variation across studies and reflects heterogeneity rather than chance. The I² statistics were used for the heterogeneity test. If the I² values were < 50%, use the fixed-effects model (FEM) to combine the effect quantity; when I² ≥ 50%, the random effect model (REM) was used for meta-analysis [18]. Independent samples test and hypothesis test summary combined the effect quantity.

To facilitate statistical analysis, percentage changes in preoperative and postoperative visual analogue scale (VAS) scores were divided by preoperative VAS scores. This generated a quantitative indicator of pain improvement. The percentage change in VAS scores was 76% (95% CI 0.76–0.82; p = 0.011, I² = 84.7%, random-effects models) for traditional groups compared with 80% (95% CI 0.78–0.82; p < 0.001, I² = 90.8%, random-effects models) for minimal groups. This outcome was no significant difference (p = 0.175) between traditional groups and minimal groups (Fig. 2A). The Funnel plot were used to assess publication bias of the change in the percentage of VAS scores after surgical removal of cysts, showing publication bias (Fig. 2B). To explore high heterogeneity, we performed sensitivity analysis using a single-study-removal method (Fig. 2C). No changes were seen in terms of the significance of outcome (Fig. 2D). Subgroup analysis of minimal groups showed no significant difference (p = 0.204) between endoscopic groups and microscopic groups, which was 79% (95% CI 0.76–0.82; p < 0.001, I² = 84.8%, random-effects models) and 81% (95% CI 0.79–0.84; p < 0.001, I² = 94%, random-effects models), respectively (Additional file 1: Fig. S4A). Publication bias was assessed by the funnel plot, suggesting publication bias (Additional file 1: Fig. S4B). When studies were excluded in the sensitivity analysis (Additional file 1: Fig. S4C), no changes were seen in terms of the significance of outcome (Additional file 1: Fig. S4D).

Analysis of the studies describing favorable last postoperative follow-up outcome using MacNab's criteria (excellent and good)/Nurick (0–2) revealed differences among traditional and minimal groups. The pooled proportion of patients experiencing a favorable outcome following excision of SSCs using traditional groups and minimal groups was 89% (95% CI 0.87–0.91; p = 0.011, I² = 59.7%, random-effects models) and 100% (95% CI 1.00–1.00; p < 0.001, I² = 75.3%, random-effects models). This outcome was significantly higher (p < 0.001) for minimal groups was than for traditional groups (Fig. 3A). Egger’s regression test yielded a p-value of 0.964, suggesting no significant publication bias (Fig. 3B). To explore high heterogeneity, we performed sensitivity analysis using a single-study-removal method (Fig. 3C). No changes were seen in terms of the significance of outcome (Fig. 3D). Subgroup analysis of endoscopic groups and microscopic groups with tubular retractors was 100% (95% CI 1.00–1.00; p < 0.001, I² = 79.0%, random-effects models) and 100% (95% CI 1.00–1.00; p < 0.001, I² = 75.1%, random-effects models), respectively. The difference between the two groups was not statistically significant (p = 0.811) (Additional file 1: Fig. S5A). Egger’s regression test yielded a p-value of 0.207, suggesting no significant publication bias (Additional file 1: Fig. S5B). To explore high heterogeneity, we performed sensitivity analysis using a single-study-removal method (Additional file 1: Fig. S5C). No changes were seen in terms of the significance of outcome (Additional file 1: Fig. S5D).

According to the meta-analysis results, the pooled proportion of dural tears in traditional and minimal groups was 0% (95% CI − 0.00–0.00; p = 0.125, I² = 32.2%, fixed-effects models) and 0% (95% CI − 0.00–0.00; p = 0.191, I² = 16.8%, fixed-effects models). The difference in dural tear between the two groups was not significant (p = 0.987) (Fig. 4A). The Funnel plots were used to assess publication bias of dural tears, suggesting publication bias (Fig. 4B). To explore publication bias, we performed trim and filling method. After adding 18 studies, the results were still not statistically significant (p = 0.976) and did not reverse (Fig. 4C). Subgroup analysis of endoscopic groups and microscopic groups with tubular retractors was 0% (95% CI − 0.00–0.00; p = 0.125, I² = 32.2%, I2 = 79.0%, fixed-effects models) and 0% (95% CI − 0.00–0.00; p = 0.306, I² = 13.7%, I2 = 23.5%, fixed-effects models), respectively. The difference between the two groups was not statistically significant (p = 0.981) (Fig. 4D). The Funnel plots were used to assess publication bias of dural tears after minimal resection of cysts, suggesting publication bias (Fig. 4E). To explore publication bias, we performed trim and filling method. After adding 12 studies, the results were still not statistically significant (p = 0.986) and did not reverse (Fig. 4F).

The meta-analysis revealed a pooled proportion of residual cyst of 0% (95% CI − 0.00–0.00; p = 0.148, I² = 29.6%, fixed-effects models) and 0% (95% CI − 0.00–0.00; p = 0.988, I² = 0%, fixed-effects models) in traditional and minimal groups, respectively. The difference in residual cyst between two groups was not significant (p = 0.994) (Fig. 5A). The Funnel plots were used to assess publication bias of residual cyst, suggesting publication bias (Fig. 5B). To explore publication bias, we performed trim and filling method. After adding 12 studies, the results were still not statistically significant (p = 0.978) and did not reverse (Fig. 5C). Subgroup analysis of endoscopic groups and microscopic groups with tubular retractors was 0% (95% CI − 0.00–0.00; p = 0.713, I² = 0%, fixed-effects models) and 0% (95% CI − 0.00–0.00; p = 0.982, I² = 0%, fixed-effects models), respectively. The difference between the two groups was not statistically significant (p = 0.988) (Fig. 5D). The Funnel plots were used to assess publication bias of residual cyst after minimal resection of cysts, suggesting publication bias (Fig. 5E). To explore publication bias, we performed trim and filling method. After adding 6 studies, the results were still not statistically significant (p = 0.984) and did not reverse (Fig. 5F).

The meta-analysis revealed a pooled proportion of recurrence of 0.07% (95% CI 0.05–0.09; p = 0.050, I² = 42.9%, fixed-effects models) and 0% (95% CI − 0.00–0.00; p = 0.984, I² = 0%, fixed-effects models) in traditional and minimal groups, respectively. The comparison of subgroup between two groups was not significant (p = 0.954) (Fig. 6A). The Funnel plots were used to assess publication bias of recurrence, suggesting publication bias (Fig. 6B). To explore publication bias, we performed trim and filling method. After adding 18 studies, the results were still not statistically significant (p = 0.978) and did not reverse (Fig. 6C). Subgroup analysis of endoscopic groups and microscopic groups with tubular retractors was 0% (95% CI − 0.00–0.00; p = 0.383, I² = 4.2%, fixed-effects models) and 0% (95% CI − 0.00–0.00; p = 0.995, I² = 0%, fixed-effects models), respectively. The difference between the two groups was not statistically significant (p = 0.983) (Fig. 6D). The Funnel plots were used to assess publication bias of recurrence after minimal resection of cysts, suggesting publication bias (Fig. 6E). To explore publication bias, we performed trim and filling method. After adding 11 studies, the results were still not statistically significant (p = 0.984) and did not reverse (Fig. 6F).

The meta-analysis revealed a pooled proportion of reoperation of 6% (95% CI 0.02–0.10; p = 0.697, I2 = 0%, fixed-effects models) and 0% (95% CI − 0.00–0.00; p = 0.007, I² = 47.1%, fixed-effects models) in traditional and minimal groups, respectively. The comparison of subgroup between two groups was significantly significant (p < 0.001) (Fig. 7A). The Funnel plots were used to assess publication bias of reoperation, suggesting publication bias (Fig. 7B). To explore publication bias, we performed trim and filling method. After adding 18 studies, the results were still not statistically significant (p < 0.001) and did not reverse (Fig. 7C). Subgroup analysis of endoscopic groups and microscopic groups with tubular retractors was 0% (95% CI − 0.00–0.00; p = 0.163, I² = 31.8%, random-effects models) and 0% (95% CI − 0.00–0.00; p = 0.005, I² = 56.4%, random-effects models), respectively. The difference between the two groups was not statistically significant (p = 0.983) (Fig. 7D). The Funnel plots were used to assess publication bias of reoperation after minimal resection of cysts, suggesting publication bias (Fig. 7E). To explore publication bias, we performed trim and filling method. After adding 11 studies, the results were still not statistically significant (p = 0.988) and did not reverse (Fig. 7F).

For surgical characteristics, independent samples test or hypothesis test summary showed a significant difference in blood loss (p = 0.027) (Fig. 8B), operation time (p = 0.361) (Fig. 8B) and postoperative length of hospital stay (p = 0.045) (Fig. 8A) between minimal and traditional groups. However, the difference in blood loss (p = 0.395) (Additional file 1: Fig. S6A), operation time (p = 0.004) (Additional file 1: Fig. S6B) and postoperative length of hospital stay (p = 0.833) (Additional file 1: Fig. S6C) between the microscopic with tubular retractors and endoscopic groups was not statistically significant.

In this review article, we pooled data from all the studies describing the results of surgical resection of LSCs to identify safer and more effective surgical management of LSCs. The pooled proportion of favorable outcomes and the percentage change were higher in the minimal groups were than in traditional groups. It is worth emphasizing that we detected publication bias by the Egger’s test on all findings and used trim and filling method to test the stability of our results. Besides, our analysis showed a significant difference between minimally and traditional groups in postoperative length of hospital stay and blood loss. We also found that minimal approaches minimized incision length, soft tissue trauma, blood loss and disruption of ligamentous and bony structures. In addition, they could produce a better clinical outcome, including complete excision of the cyst, decreased postoperative pain and reduced hospitalization period, and were more cost-effective than traditional surgery. Interestingly, we found that the probability of reoperation in the minimal groups was lower than that in the traditional groups. Moreover, minimally invasive microscopic approaches with tubular retractors are considered far more challenging than conventional approaches due to a high risk of the residual cyst, dural tear and recurrence [19‐21]. However, our analysis revealed no significant difference in risk of dural tear, residual cyst, recurrence, and reoperation between minimal and traditional groups. Compared to microscopic excision of LSCs, endoscopic groups with tubular have less operative time. Other analyses revealed no statistically significant difference between subgroups. However, we should also consider operative time (with the associated learning curve involved for surgeons) and accessibility to theater and equipment to perform this specialized surgery [22‐25]. In order to better understand this technology, we have made a tabulated summary of the key findings and also comparative literature of the endoscopic approach to other lesions (Additional file 1: Table S1) [25‐28].

Recent studies reviewed described spontaneous resolution after cyst rupture, triggering a local inflammatory reaction because of the presence of prostaglandins, proteases, and cytokines [29‐32]. Because of this phenomenon, cysts strongly adhere to the dura mater due to intermittent small ruptures, hindering cyst resection and increasing the risk of incidental durotomy. Most scholars believe that the traditional surgical approach is the safest approach for excision of synovial cysts, whether decompression or fusion [11, 27]. It seems reasonable to consider that the microscopic with the tubular approach is far more challenging than the conventional one because of risk of epidural hematoma, durotomy and cerebrospinal fluid leak, especially in narrower epidural spaces [12]. However, many studies have reported safe removal of spinal canal synovial cysts using minimally invasive microscopic approaches with tubular retractors and endoscopic approaches. Similarly, many studies have reported using endoscopic approaches and minimally invasive microscopic approaches with tubular retractors in the treatment of LSCs, demonstrating that LSCs can be safely resected with good outcomes [33‐35]. In our study, of the 38 patients with incidental dural tears during minimally invasive tubular approach, 88% were managed conservatively or through primary repair.

In the meta-analysis, joint destabilization was one of the causes of recurrent intraspinal synovial cyst, specifically spondylolisthesis. The presence of spondylolisthesis varied between 23 and 88% (mean, 31.5%) [36]. In our study, we obtained a similar result. Of the 1601 patients, 31% had preoperative spondylolisthesis, supporting, fusion as a first therapeutic choice in most cases [37]. Using minimally invasive surgery, Rolemberg, Scholz and Denis all found no difference in recurrence of radiculopathy, back pain and cyst between patients with decompression alone and those with decompression and fusion [4, 37, 38]. Gupta et al. considered that fusion surgery prolongs surgery and, overall, has more surgery-related risks than sole decompression of SSCs. It also presents additional problems including screw loosening, adjacent level degeneration or breakage [39]. In our meta-analysis, in traditional groups, 31% of patients had preoperative spondylolisthesis, whereas 15.7% underwent fusion. In minimal groups, 27.24% of patients had preoperative spondylolisthesis, whereas 4.02% underwent fusion. Lastly, in the endoscopic groups, 18.64% of patients had preoperative spondylolisthesis, and none underwent fusion. There was no difference in recurrence among the three groups and minimal groups had better functional improvement and less reoperation rates. The analysis results also proved that decompression alone was enough to achieve good results in preoperative stable grade 1 spondylolisthesis patients and low incidence of secondary fusion surgery. Our analysis revealed no difference in postoperative functional recovery, cyst recurrence and reoperation between decompression alone group and decompression with fusion group showed (p > 0.05). This conclusion is inconsistent with Khan et al. [40]. A possible explanation for this discrepancy is that their study probably compared only the curative effect difference between fusion and without fusion surgery and ignored the impact of surgical methods [41‐43]. Minimally invasive surgery preserves small joints, reduces damage to the surrounding muscles, bones and ligaments, and prevents iatrogenic segmental instability to the greatest extent, especially in preexisting spondylolisthesis [44, 45].

Of the two cases we included in our study, case 1 was a patient with lung metastases from colorectal cancer and case 2 was a patient with chronic myeloid leukemia. Although the two patients had longer intraoperative time due to underlying comorbidities, they recovered well after surgery. Case 2 was hospitalized for a long time due to comorbid fever and high inflammatory indexes after operation. However, on the first day after the operation, he could go to the ground and preoperative low back pain and radiculopathy had been relieved. In our experience, minimally invasive surgery for spinal synovial cysts is characterized by less intraoperative bleeding and damage to the surrounding muscles, bones and ligaments, hence, a better option for patients with severe underlying diseases.