This review quantitatively synthesized the currently available evidence on the prevalence and risk factors associated with the incidence of postoperative CSF leakage after TSS in 34 studies, which included 9,144 participants with PA, to demonstrate a pooled global prevalence of 5.6%. We evaluated the relationships between these nine factors and the risk of postoperative CSF leakage. Our data indicated that tumor size, adenoma consistency, revision surgery, and intraoperative CSF leakage were independent risk factors for postoperative CSF leakage. By contrast, the use of an endoscopic approach showed a significant protective benefit compared with the microscopic approach in TSS, whereas sex, functional adenoma type, resection rate, and perioperative LD were not found to be related to the occurrence of postoperative CSF leakage.
The effect of tumor size on postoperative CSF leakage is under debate, with some studies showing that patients with giant adenoma are more likely to experience postoperative CSF leakage than those with macroadenoma or microadenoma[8, 10]. In contrast, Shiley et al.[46] and Nishioka et al.[47] found that CSF leakage was more common among patients with microadenomas. In addition, Chi et al.[48] reported that tumor size was not associated with postoperative CSF leakage. Han et al.[10] indicated that large pituitary tumors commonly expand the sella and erode regions adjacent to the meninges, resulting in the attenuation of these barriers to the CSF space. Moreover, larger pituitary tumors result in the wider invasion of adjacent tissues, resulting in a greater extent of resection, which may increase the risk of CSF leakage after TSS[19]. In our study, tumor size was identified as a risk factor for postoperative CSF leakage, with high heterogeneity. Therefore, more high-quality research is necessary to better explore the effects of tumor size on postoperative CSF leakage occurrence.
The current analysis showed that adenoma consistency was associated with the occurrence of postoperative CSF leakage. Most PAs have a soft texture that can easily be resected with curettage and suction[49]. Hard (fibrous) tumors account for approximately 5–13% of PA and are difficult to separate from critical structures, often requiring removal using sharp dissection or laser techniques[10, 49]. Therefore, the consistency of an adenoma affects its resection success[50]. In our meta-analysis, compared with soft tumors, firm pituitary tumors were found to be associated with a greater risk of CSF leakage after TSS, with crude ORs of 2.13–4.81. Therefore, the PA consistency must be assessed using reliable imaging techniques (such as magnetic resonance imaging, MRI) before performing TSS, which may help surgeons to better plan an appropriate operative strategy and reduce the risk of surgery.
A strong association was observed between revision surgery and an increased risk of postoperative CSF leakage, which is consistent with the findings of a previous study[10, 46, 47]. Shiley et al.[46] found that the incidence of postoperative CSF leakage after revision surgery was significantly higher than that after primary surgery (14.6% vs. 4.0%, P = 0.010) in a study examining 235 patients undergoing TSS. The first procedure can create scar tissue that adheres to the arachnoid space and diaphragm[51]. The revision surgery is made more difficult by the presence of adhesions, tissue fibrosis, abnormal vascular hyperplasia, and distorted anatomy, which can increase the complexity of the dissection [20]. Moreover, revision surgery requires the removal of residual tumors that were not easily detected or resectable during the prior surgery, which typically requires a more aggressive dissection approach along the sellar diaphragm, increasing the risk of postoperative CSF leakage[20].
In our results, patients with an intraoperative CSF leak were 6.33 times more likely to experience postoperative CSF leakage than patients without intraoperative CSF leakage. Seiler et al.[52] found that the occurrence of postoperative CSF leakage was six times as common among cases that reported intraoperative CSF leakage compared with those that did not. Similar results were observed in a retrospective analysis, which showed a significant difference in the rates of postoperative CSF leakage between patients who experienced an intraoperative CSF leak and those who did not (16.7% vs. 2.3%)[10]. Not surprisingly, intraoperative CSF leakage was correlated with the risk of postoperative CSF leakage, which may be due to the incomplete repair of intraoperative CSF leakage[20]. The unidentified or delayed development of intraoperative CSF leaks is also an equally important source of postoperative CSF leakage as the failure to employ effective CSF leak repair methods[53]. In addition, repair materials may shift or fall off due to postoperative actions that increase intracranial pressure (such as sneezing and constipation), resulting in incomplete leakage closure. Sarita et al.[54] found that chronic cough was one of the primary contributing factors to the failure to resolve an intraoperative leak, leading to the postoperative recurrence of the leak. Our findings suggested that patients with intraoperative CSF leaks may warrant more aggressive management to prevent the development of postoperative leaks, and correct behavior education is also important for the management of postoperative CSF leakage.
PAs with high levels of invasiveness are more likely to invade the peritumoral tissues, such as the cavernous sinus and the internal carotid artery, which may increase the difficulty of TSS. To completely remove the tumor, the surgeon may overstretch the sellar septum and dura mater, which increases the risk of CSF leakage[14]. In addition, the arachnoid membrane at the sellar diaphragmatic foramen can be easily damaged when most of the tissue around the PA is removed, resulting in CSF leakage. In our meta-analysis, few studies focused on the effects of invasiveness of PA on the occurrence of postoperative CSF leakage; therefore, we did not confirm whether this variable may serve as a clinically useful marker of an individual’s propensity toward postoperative CSF leakage. However, in cases of PA with high invasiveness, more attention should be given to local anatomy and surgical skill[55]. To prevent the occurrence of CSF leakage during TSS, the PA should be removed according to appropriate procedures, and the traction of the peritumoral tissues should be reduced as much as possible to avoid causing damage to the sellar septum and dura mater.
The current study found a mild association between the performance of endoscopic TSS and a reduced risk of postoperative CSF leakage, with crude ORs of 0.50–0.96. A meta-analysis of 23 observational studies conducted by Li et al.[56] previously showed that endoscopic TSS had no significant effect on the risk of CSF leakage compared with microscopic TSS. However, the data analyzed by Li et al.[56] cannot be directly compared with our series because their results were not limited to postoperative CSF leakage and encompassed all pituitary pathology. In addition, relatively few complications were reported. Compared with the microscopic approach, the endoscopic approach has many advantages. It is easier, requires a less traumatic entry into the sphenoid sinus, and enables wide and close views of the tumor, allowing for increased tumor resection[37]. By contrast, some surgeons believe that the endoscopic approach decreases stereoscopic visualization and decreases the ability to use their instruments[57], which may account for the lack of significant difference observed for the incidence of CSF leakage between these two surgical techniques in some studies[34, 37]. In our study, the endoscopic approach showed a minimally protective benefit for reducing postoperative CSF leakage compared with the microscopic approach. Surgeons must be specially trained for an endoscopic approach[37], and the surgeon’s learning curve was found to be associated with the occurrence of postoperative CSF leaks[58]. The results of endoscopic TSS may improve with the experience of individual surgeons. Therefore, further research remains necessary to explore the potentially protective role of using the endoscopic approach to prevent the occurrence of postoperative CSF leakage.
Although sex, aging, and body mass index (BMI) were reportedly associated with postoperative CSF leakage[58], the current evidence could not be used to determine whether these demographic factors were risk factors. A multi-institutional study of patients undergoing endoscopic PA showed that the risk of postoperative CSF leakage in female patients was 2.4 times higher than that in male patients[59]. However, Zhang et al.[23] and Tian et al.[20] have shown that the female sex of patients had no effect on the occurrence of postoperative CSF leakage. Whether sex is a risk factor for postoperative CSF leakage requires further investigation. Relatively few studies have examined the impacts of aging and BMI on postoperative CSF leakage in PA patients undergoing TSS, and various definitions of aging and BMI were used in these studies, making the meta-analysis of these two factors difficult to perform. Caitlin et al.[59] found that younger patients (< 64 years) had a higher risk of postoperative CSF leakage than those older than 64 years. In two retrospective analyses, no correlation was observed between average patient age and the occurrence of postoperative CSF leakage[46, 47]. Several studies showed that an elevated BMI was an independent predictor of postoperative CSF leakage after TSS[59, 60]. This association might be due to the increased intra-abdominal pressure associated with higher BMI [58]. A moderate level of physical activity (e.g., 150 minutes of moderate aerobic exercise) for BMI reduction may be helpful[61]. In addition, more studies remain necessary to quantify the effects of aging and BMI on the occurrence of postoperative CSF leakage in PA patients undergoing TSS.
The present meta-analysis could not determine whether the functional adenoma type was associated with an increased risk of postoperative CSF leakage in PA patients undergoing TSS. Two studies found that postoperative CSF leakage was more common in adrenocorticotropic hormone (ACTH)-producing adenomas[46, 47], which is likely due to ACTH adenomas often not appearing localized on preoperative imaging and requiring a more aggressive resection approach[46, 47]. However, only a few events of postoperative CSF leakage were reported in these studies, and the conclusions were prone to bias. Han et al.[46, 47] found that the postoperative leakage rates associated with follicle-stimulating hormone (FSH) adenomas were higher than those for other tumor types. By contrast, Tian et al.[22] and Wang et al.[22] found that the functional adenoma types were not a significant risk factor for postoperative CSF leakage; therefore, this inconsistency requires further study.
Whether the resection rate is a risk factor for postoperative CSF leakage remains controversial. A retrospective analysis of 1,641 patients with PA undergoing TSS showed that the incidence of postoperative CSF leakage in patients with total resection was significantly higher than that among patients with partial or subtotal resections[14]. The mechanism underlying this association may be that tumors that undergo total resection, especially for giant or invasive adenomas, experience more severe dural stretch, increasing the risk of postoperative CSF leakage[14]. The literature exploring the resection rate remains scarce and reported results have been conflicting and highly heterogeneous. Further studies must be performed to confirm the relationship between the resection rate and the occurrence of postoperative CSF leakage.
The use of LD remains under debate. In a meta-analysis[5], LD had an OR 1.13 for reducing the occurrence of postoperative CSF leakage after endonasal endoscopic skull base surgery. Gautam et al.[40] posited that preoperative or intraoperative LD could reduce tension on the arachnoid caused by the expansion of a pituitary macroadenoma, preventing the potential exposure of the arachnoid to intraoperative injury and reducing the risk of intraoperative CSF leakage. However, Jung et al.[5] suggested that these results should be interpreted carefully because surgeons tend to perform LD in high-risk patients or when the surgeon feels that the reconstruction was not completely successful. A prospective study of 114 TSS procedures for pituitary macroadenoma indicated that LD reduced the rate of intraoperative CSF leakage from 41–5% (p < 0.001), but the rate of postoperative CSF leakage remained similar (5% vs. 5%). In addition, some researchers have stated that the presence of the drain may mask the earlier detection of a CSF leak[43]. Therefore, whether LD reduces the risk of postoperative CSF leakage after TSS remains inconclusive given the current evidence.
The acknowledged limitations of this study should be mentioned. First, unpublished articles were not included in this systematic review, likely contributing to publication bias. Second, a lack of randomized control trials (RCTs) examining the occurrence of postoperative CSF leakage exists, which may reduce the reliability of the results of our study. Further studies using RCT designs may enrich and substantiate our results. Despite these limitations, this study represents the first known meta-analysis to examine the risk factors associated with postoperative CSF leakage after TSS in PA patients. This study makes a significant contribution to clinical practice because our findings can focus clinical medical workers’ attention on the occurrence of postoperative CSF leakage after TSS in PA patients and provide foundations for targeted treatment strategies.